\documentclass[11pt]{article}
    
    
\usepackage{amsmath} 
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{booktabs}
\usepackage{bm}
\usepackage[usenames, dvipsnames]{color} 
\usepackage{dsfont}
\usepackage{graphicx}
\usepackage{hyperref}
\usepackage[utf8]{inputenc}
\usepackage{lscape}
\usepackage{natbib}
\usepackage{multirow}
\usepackage{setspace}
\usepackage[capposition=top]{floatrow}
\usepackage[margin=0.9in]{geometry}
\usepackage{subfloat}
\usepackage{caption}
\usepackage{xr}
\usepackage[caption=false]{subfig}
\usepackage{sectsty}
\usepackage{titlesec}
\usepackage{titletoc}
\usepackage[bottom,hang,flushmargin]{footmisc}
\usepackage{longtable}
\usepackage{colortbl}

\sectionfont{\fontsize{14}{15}\selectfont}
\subsectionfont{\fontsize{13}{15}\selectfont}
\titlespacing\section{0pt}{12pt plus 0pt minus 0pt}{0pt plus 0pt minus 0pt}
\titlespacing\subsection{0pt}{12pt plus 0pt minus 0pt}{0pt plus 0pt minus 0pt}



\setlength\parindent{0.25in}
\setlength\parskip{0.1in}

\hypersetup{        
    colorlinks=true,   
    linkcolor=black,
    citecolor=black,
    filecolor=black,
    urlcolor=blue
}

\usepackage{fontspec}
\setmainfont{Times New Roman}

%-------------------------------------------------------------------------------
\begin{document}
\title{\textbf{Maternal Mortality and Women's Political Power}\thanks{Bhalotra: Department of Economics, University of Warwick, United Kingdom, CEPR, AcSS, IZA, IEA, CAGE, MiSOC \href{mailto:sonia.bhalotra@warwick.ac.uk}{sonia.bhalotra@warwick.ac.uk}; Clarke: Department of Economics, Universidad de Chile, MIPP, and IZA,  \href{mailto:dclarke@fen.uchile.cl}{dclarke@fen.uchile.cl}; Gomes: IRES/LIDAM, UCLouvain Belgium and CEPR, \href{mailto:joseph.gomes@uclouvain.be}{joseph.gomes@uclouvain.be}; Venkataramani: University of Pennsylvania and NBER, \href{mailto:atheenv@pennmedicine.upenn.edu}{atheenv@pennmedicine.upenn.edu}. We acknowledge support of UNU-WIDER in Helsinki who commissioned an earlier version of this paper in connection with the 17\textsuperscript{th} Nordic Conference on Development Economics. We are grateful to James Fenske, Alessandro Tarozzi, Mark Pauly, Dean Yang, the editor Paola Giuliano and anonymous referees for comments. 
Replication materials are available at \url{https://doi.org/10.7910/DVN/INLWUH}. Bhalotra acknowledges partial support for her time from CAGE and the Department in Warwick and from ESRC grant ESM010236-1 awarded to the Human Rights, Big Data and Technology project at the Human Rights Centre in Essex. Clarke acknowledges support from FONDECYT Chile grant 1200634, and from the Institute for Research in Market Imperfections and Public Policy, MIPP.}} %(ICS13\_002 ANID).}}
\author{Sonia Bhalotra\\ \normalsize{University of Warwick}
  \and Damian Clarke\\ \normalsize{Universidad de Chile}
  \and Joseph F. Gomes\\ \normalsize{UCLouvain}\\\\
  \and Atheendar Venkataramani\\ \normalsize{University of Pennsylvania}}

\date{\today}

\renewcommand{\thefootnote}{\arabic{footnote}}
\renewcommand{\footnotesize}{\small} 
\setcounter{footnote}{0} 
\thispagestyle{empty}
\maketitle

\begin{spacing}{1.2}
\begin{abstract} 
Millions of women continue to die during and soon after childbirth, even where the knowledge and resources to avoid this are available. We posit that raising the share of women in parliament can trigger action. Leveraging the timing of gender quota legislation across developing countries, we identify sharp sustained reductions of 7--12 percent in maternal mortality. Investigating mechanisms, we find that gender quotas lead to increases in percentage points of 5--8 in skilled birth attendance and 4--8 in prenatal care utilization, alongside a decline in fertility of 6--7 percent and an increase in the schooling of young women of about 0.5  years.  The results are robust to numerous robustness checks. They suggest a new policy tool for tackling maternal mortality.
\end{abstract}

\noindent\emph{JEL codes:} I14, I15, O15.\\
\emph{Keywords:} maternal mortality, women's political representation, gender, quotas, reproductive health services, fertility, schooling.
    
\thispagestyle{empty}
\setlength{\baselineskip}{1.3\baselineskip} 
\newpage 
\end{spacing}
\begin{spacing}{1.3}

\section{Introduction}
Maternal mortality, defined as the death of women within 42 days of childbirth, remains a looming global health problem well into the 21\textsuperscript{st} century. It is estimated to account for 830 deaths per day and more than 216 deaths per 100,000 live births globally \citep{ceschia2016maternal}. In sub-Saharan Africa, the maternal mortality ratio (MMR) exceeds the rate in developed countries a century ago \citep{Alkemaetal2016,loudon1992death}.\footnote{MMR is defined as maternal deaths per 100,000 live births. In sub-Saharan Africa in 2015 it was 547; in the US in 1936 it was 555.} A woman's lifetime risk of maternal death (the probability that a 15 year old woman will eventually die from a maternal cause) is 1 in 190 today, but there are dramatic variations across the world, the risk being 1 in 5400 in high income countries and 1 in 45 in low income countries \citep{WHO2019}. Moreover, maternal mortality is only the tip of an iceberg, the mass of which is maternal morbidity \citep{koblinsky2012maternal}.\footnote{For every woman who dies from obstetric complications, approximately 30 more suffer injuries, infection and disabilities \citep{hunt2007reducing}. In 1999, for example, the WHO estimated that over 2 million women living in developing countries remain untreated for obstetric fistula, a devastating injury of childbirth. Of the 6 million missing women each year, 21\% are in their reproductive years \citep{wong2012world}.} There is no single cause of death and disability for men aged 15--44 that is close in magnitude. 

Reducing maternal mortality is of both intrinsic and functional value, as it favorably influences women's human capital attainment, employment, and growth \citep{AlbanesiOlivetti2016,albanesi2014maternal,JayachandranLlerasMuney2009,Bloometal2015}. A broad stream of research has documented the importance of population health for economic growth, via life expectancy and human capital accumulation \citep{soares2005mortality,weil2005accounting,ashraf2008does,shastry2003much,bloom2004effect,lorentzen2008death,aghion2010relationship}.
    
Persistence of high rates of maternal mortality is striking given that the knowledge and technology needed to dramatically reduce it have been available for nearly a century, and the costs of intervention are relatively small \citep{cutler2006determinants,loudon1992death}. The causes of maternal mortality are well understood, and have not varied a lot through the course of history. Skilled care before, during and after childbirth can prevent about three-fourths of maternal deaths \citep{WHO2019,hunt2007reducing}.  Rather than obstetricians and gynecologists, this requires relatively low-cost primary care during pregnancy and midwives at delivery \citep{BhalotraRochaSoares2019,pettersson2014midwives,Tikkanenetal2020}. In recognition of this, the Millennium Development Goals (MDG) set in the year 2000, included as a target for 2015 universal access to reproductive health services. Progress was made, but it fell short of this target \citep{zureick2013understanding}. 

As more than 95\% of maternal mortality occurs in developing countries, a natural explanation may be that low income has constrained progress. However, there is considerable variation in levels and in rates of decline of MMR conditional upon income, see for example \cite{Ritchie2020Exemplar}. Among low income countries, MMR in Rwanda is three times lower than in Chad despite it being poorer. Among high income countries, the United States has the highest MMR despite its considerable wealth. Bucking the global tide, the US has seen an increase in MMR of about 50\% since 2000 \citep{macdorman2016united,mann2018we}.Although aggregate income displays a positive association with each of female and male life expectancy, it exhibits only a weak relationship with the ratio of female to male life expectancy, a proxy for excess deaths of women associated with reproduction (Appendix Figure \ref{fig:LELnGDP}). Overall, it seems that other factors are at play.

We put forward the hypothesis that the paucity of women policy-makers has constrained progress. In particular, we argue that male-dominated parliaments have not sufficiently prioritized maternal mortality reduction. This may reflect both preferences and information constraints. Women leaders may be innately more concerned about MMR because they identify with the risks, or have clearer information on the risks \citep{ashraf2020maternal,powley2007rwanda}.
The broad stylized facts line up with our hypothesis: since 1990, MMR has shown an unprecedented fall of 44\%, a period in which the share of women in parliament has risen unusually rapidly, from under 10\% to more than 20\% (Figure \ref{fig:rawMMR}). We study whether these trends are causally related. 


To do this, we leverage the abrupt legislation of reserved seat quotas for women in parliament. A wave of quota adoption swept through developing countries, with 22 countries reforming during 1990--2015, the period we analyse. We first demonstrate that quota legislation leads to a sharp increase in the share of parliamentary seats held by women. We then show that quota adoption is associated with an 8--12\% decline in the maternal mortality ratio (MMR). The dynamic impacts are persistent, in fact increasing over time, consistent with women who enter parliament after quotas remaining through a five-year term, and with the efforts of parliamentarians cumulating over time. Quota impacts are increasing in the pre-intervention level of MMR, consistent with the low hanging fruit argument. They are also increasing in the share of seats reserved, and finding this dose-response relationship increases our confidence that we capture impacts of quota adoption rather than a confounder.

We use data on maternal mortality that were recently released by the WHO \citep{Alkemaetal2016,Alkemaetal2015}, merged with data on gender quota legislation, the share of women in parliament and a host of other data that allow us to analyse mechanisms and confounders.
The main analysis relies upon the \cite{dCDH2020} estimator which produces unbiased estimates of dynamic effects when treatment effects are heterogeneous across units or time, avoiding the potential bias arising in conventional difference-in-differences estimates.
The identifying assumption is that the country-specific timing of quota adoption is quasi-random. The main threat to identification is that an underlying trend in gender progressivity may have triggered quota legislation and also led to lower MMR, with the quotas having no causal impact on MMR. This would be in line with broader evidence that the timing of legislation is not random but, instead, that legislation is passed when social preferences have evolved to support it \citep{DoepkeZilibotti2005,PLATTEAU2014633}.

We probe this and other threats to identification. 
The placebo coefficients show no evidence of differential pre-trends. We show that the main results hold if we use estimators that make weaker assumptions about parallel trends, including a standard synthetic control approach \citep{Abadieetal2010} and the synthetic difference-in-differences approach \citep{arkhangelsky2021synthetic} and, following \citet{RambachanRoth2019}, we estimate bounds on the dynamic effects after relaxing the parallel trends assumption. 
We further estimate a limited information maximum likelihood (LIML) model in which MMR is regressed on the share of women in parliament, instrumented with quota legislation. Following \citet{Conleyetal2012}, we estimate bounds on these estimates, allowing that the exclusion restriction is violated, as would be the case if quota adoption had a direct impact on MMR conditional on the share of women in parliament because quota adoption proxies an underlying move towards gender equality. These results show that increasing the share of women in parliament by 1 percentage point (pp) generates a drop in MMR of 1.5--2\%. Estimates of Anderson-Rubin confidence sets show that this finding holds even after accounting for the instrument being weak. We demonstrate robustness of the estimates to inclusion of country-specific linear trends and region$\times$year fixed effects. 

We directly investigate the evolution of a policy environment supportive of gender equality. We find no evidence that quota adoption is preceded by an uptick in any of 18 distinct indicators of women’s economic rights, civil liberties and property rights. We nevertheless also confirm that our estimates are not sensitive to conditioning upon these indicators. Our findings are not entirely surprising -- while the progression of gender equality is likely to eventually culminate in increased attention to women's reproductive health, this is likely to be a slow process. In contrast, quotas that give women \emph{instrumental} power to action improvements can produce sharp changes. We demonstrate this, showing that the decline in MMR following quota adoption is mirrored in improved coverage of reproductive health services and movements in other relevant mechanisms.

We show that our results are not driven by democratization, nor by other drivers of quota adoption identified in the political science literature \citep{Krook2009}. To further support identification and affirm the broad generality of our findings, we leverage the quasi-random state-level adoption of gender quotas in local government in India \citep{Iyeretal2012}, where we identify a significant relationship of broadly similar magnitude.


As the main analysis uses aggregate data, we confirm that our findings are not driven by changes in the composition of women giving birth. To do this we created a pseudo panel from the full reproductive histories of women available in harmonized microdata including 10,837,442 births of 3,079,298 individual women surveyed in 34 different years in 82 countries, 14 of which implemented quotas. 
Using three different approaches, we show that our estimates are robust to accounting for uncertainty in the MMR data.

Turning to mechanisms, we identify quota-led improvements in reproductive health services coverage, the key determinant of MMR highlighted in scientific research and promoted by World Health Organization (WHO) guidelines \citep{grepin2013maternal,kruk2016next}. We identify increases in coverage of skilled birth attendance of 5 to 8 percentage points (6--9.6\%), of antenatal care of 4 to 8 percentage points (4.8--9.6\%) and an increase in contraceptive coverage of 1.7 percent. However, we find no impact on adoption of abortion legislation (relevant because unsafe abortion is a major cause of MMR \citep{girum2017correlates,ClarkeMuhlrad2021}). This is possibly because there are religious barriers to abortion law which women leaders may not be able to surmount.

We also look at whether an increased presence of women in parliament modified demand side determinants of MMR, including women's rights, female education and fertility \citep{BhalotraClarke2013,girum2017correlates}. We find no discernible impacts of quotas on women's rights but we identify an increase of 0.5 years in the education of young women and a decline in the total fertility rate of 6--7\%, consistent with the observed expansion of contraceptive coverage and women's schooling. In addition to being a likely mechanism for the decline in maternal death risk per birth (the definition of MMR), the decline in fertility will have a \emph{scale effect}, tending to reduce the number of maternal deaths at any level of risk per birth. We estimate that the scale effect leads to an additional decline in the maternal death count that is roughly 64\% of the impact of quotas on maternal deaths per birth.\footnote{Our back-of-the-envelope calculations suggest that more than 8000 maternal deaths a year were averted by quotas lowering maternal deaths per birth and an additional 5669 maternal deaths a year were averted on account of the impact of quotas on the number of births.} 

In view of broader concerns about reducing MMR in low-resource settings, a natural question is whether these results rely upon women parliamentarians raising resources. We find no evidence that gender quotas arrest declines in male reproductive-age mortality, mortality from tuberculosis (which is gender-neutral) or infant mortality -- thus no evidence of substitution. We find some evidence of an increase in health spending, albeit this is sensitive to specification. We find no evidence that more resources became available -- there is no impact of gender quotas on GDP or international development assistance for maternal health (DAH). We argue that large changes in outcomes are feasible without a large increase in resources, highlighting the low costs of relevant interventions \citep{banke2020cost,Dupas2011}, and the slack generated by inefficiency and corruption that a motivated leader can put to use \citep{Folbre2012,BrolloTroiano2016,Baskaranetal2018}. In section \ref{scn:mechanisms}, we cite evidence consistent with leaders being able to move the outcomes they prioritize by building consensus, provoking legislation and improving policy design and delivery, including better targeting and greater outreach. 

Overall, we provide evidence that the political empowerment of women can effect rapid maternal mortality decline. Thus gender quotas may be a powerful at-scale means of modifying policy priorities in favor of maternal health. 

The rest of this paper is organized as follows. In section \ref{scn:background} we provide a discussion of maternal mortality initiatives and of the implementation of parliamentary gender quotas. Section \ref{scn:data} describes the data, and section \ref{scn:ES} lays out the empirical strategy. We present results in section \ref{scn:results}, robustness checks in section \ref{sscn:robustExtend} and analysis of mechanisms in sections \ref{scn:mechanisms} and \ref{scn:resources}. We conclude in section \ref{scn:conclusions}.  
    

\section{The Policy Landscape}
\label{scn:background}
\noindent\textbf{Maternal Mortality.} Although women have been dying in and around childbirth since the origin of life, international initiatives to tackle maternal mortality are recent.  The Safe Motherhood Conference held in 1987 in Nairobi was the first of a series of international meetings that highlighted the need for global action on maternal mortality. Strategies for achieving this goal included making family planning universally available, providing prenatal care and trained assistance at delivery, and ensuring access to emergency obstetric care \citep{starrs2006safe}. Subsequent events calling for action include the World Summit for Children in 1990 and the International Conference on Population and Development in 1994. In September 2000, the United Nations General Assembly adopted the UN Millennium Declaration and articulated the Millennium Development Goals (MDGs). MDG 5 called for a three-quarters reduction in MMR between 1990 and 2015. A second target of achieving universal access to reproductive health by 2015 was added in a subsequent reformulation of the MDGs. The Sustainable Development Goals agreed in 2015 set new, more ambitious targets to be achieved by 2030. However, no new policies to hasten progress have been proposed.



\noindent\textbf{Gender Quotas.} In response to growing awareness of women's rights in civil society, in 1990 the UN Economic and Social Council set a target of 30\% female representation in decision making bodies by 1995 \citep{PandeFord2012}. The passage of gender quotas followed this and accelerated after the unanimous signing of the Beijing Platform for Action by all UN delegates at the Fourth World Conference on Women in 1995 \citep{IPU2015,Krook2009}. Since 1990, 22 countries in sub-Saharan Africa, the Middle East, and South and East Asia have implemented constitutionally protected quotas reserving seats in parliament for women, mostly after 1995. We observe an uptick in quotas particularly after year 2000, driven by sub-Saharan Africa.\footnote{The countries implementing quotas are: Afghanistan, Algeria, Bangladesh, Burundi, China, Djibouti, Eritrea, Haiti, Iraq, Jordan, Kenya, Morocco, Niger, Pakistan, Rwanda, Saudi Arabia, South Sudan, Sudan, Swaziland, Tanzania, Uganda and Zimbabwe. Uganda is the only country which reserved seats before 1990, in 1989. Four other countries: Samoa, Kosovo, Somalia and Taiwan have implemented quotas during 1990--2015, but are left out due to data restrictions. Three other countries implemented quotas more recently and fall outside our study period: Nepal in 2016, UAE in 2019 and Egypt in 2020.} 
The existing evidence on reserved seat quotas is dominated by analysis of the randomized implementation of gender quotas (for headship together with council membership) in local government in India.\footnote{In their pioneering study, \citet{chattopadhyay2004women} showed that women leaders were responsive to the needs of women citizens, as expressed in council meetings. Direct observation of council meeting minutes confirms this level of interaction \citep{Parthasarathy2019DeliberativeDI}, also see \citet{Iyeretal2012}.} We also provide estimates for the Indian quotas, alongside our analysis of women's membership of parliament. 

The focus of our study is on these reserved seat quotas but we shall provide a brief analysis of impacts of candidate quotas on women's share in parliament and MMR. Since 1990 the number of countries with candidate list quotas for women has also risen sharply, from 1 to 46.\footnote{Candidate quotas set a minimum for the share of women on candidate lists, either as a legal requirement or a measure written into the statutes of individual political parties.} Candidate quotas were passed mostly in middle- and higher-income countries (see Appendix Figure \ref{quotacover}) and have weaker impacts on representation, see for example, \cite{BaguesCampa2017}.

\section{Data and Descriptive Statistics} 
\label{scn:data}
\noindent\textbf{Maternal Mortality.} Maternal mortality was not consistently measured until recently, imposing an impediment to evidence-based prevention efforts. The MDGs set quantitative targets to be monitored, and this triggered a multi-agency effort to gather data on MMR. In this paper we use the first harmonized time series estimates of MMR across 183 countries, released in 2015 by the United Nations Maternal Mortality Estimation Inter-Agency Group (MMEIG), covering the period 1990--2015. These estimates combine available data from vital statistics, special inquiries, surveillance sites, population-based household surveys, and census files. They use Bayesian methods to combine these data and fill gaps \citep{Alkemaetal2016,Alkemaetal2015}. We conduct a sensitivity check that allows for this in inference, and we also show results using MMR data derived from the Demographic and Health Surveys (DHS). 
    
\noindent\textbf{Gender Quotas and Women's Share in Parliament.} We collected information on country-specific adoption of quotas up until 2005 from \citet{Dahlerup2005}, and updated these to 2015 using the Global Database of Quotas for Women. The data include the date of legislation and the share of seats reserved for women. We obtained the share of women in parliament from the World Development Indicators (WDI), the UN Millennium Development Goals (MDG) Indicators and the ICPSR dataset compiled by \citet{Paxtonetal2008}. Figure \ref{fig:rawWP} shows that aggregate trends in women's share in parliament track trends in quota coverage. 
    
\noindent\textbf{Other Data.} Data on a range of intermediate outcomes (mechanisms) and controls, including measures of women's rights or gender equality, political variables and indicators of reproductive health coverage were compiled from diverse sources, see Appendix \ref{app:Data}, where we also discuss the MMR and micro-fertility data drawn from the DHS, and state-level  quota adoption, women in government, and MMR data for India. 

\noindent\textbf{Descriptive Statistics.} The analysis sample contains (at most) 178 countries, through 1990--2015. Table \ref{ssquota} provides summary statistics. Appendix Figure \ref{MMRworld} plots the world distribution of average MMR in the analysis period. The geographic spread of reserved seat quotas is in Figure \ref{quotacover} and the trend in gender quota implementation in Figure \ref{quotatiming}. Quota size varied across countries and Figure \ref{sizeQuotas} displays the  distribution. The median (mean) gender quota is 21\% (20\%). Casual inspection suggests support for our hypothesis that reserved seat quotas are associated with MMR decline. Comparing country pairs with similar GDP per capita in 1990, selecting one which implemented reserved seat quotas before 2010 and one which did not, we find that the quota-implementing country tends to witness a larger decline in maternal mortality in 1990--2010. Thus, Burundi did better than Malawi, Kenya did better than Zimbabwe, and Niger did better than the DRC. A more formal approach is discussed next.
    
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    
\section{Empirical Strategy}
\label{scn:ES}
The share of women in parliament has increased fairly smoothly (Figure \ref{fig:rawMMR}), making it hard to isolate its effects from those of other gradually evolving trends. We therefore leverage the abrupt implementation of quotas. The main results are obtained using the \cite{dCDH2020} estimator. However, we also provide the standard event study estimates, the Goodman-Bacon decomposition, bounds following \citet{RambachanRoth2019} and estimates based on synthetic controls \citep{Abadieetal2010,arkhangelsky2021synthetic}. We begin by setting out the standard event study model \citep{Jacobsonetal1993}, that allows us to track outcome trends before and after quota implementation. The estimated equation is:
\begin{eqnarray}
  \label{eqn:wp-q}
  Y_{ct} &=&\alpha + \sum_{l=2}^{10+} \beta_{l}^{lead}Quota_c\times 1\{lead_t=l\} + \sum_{k=0}^{10+} \beta_{k}^{lag}Quota_c\times 1\{lag_t=k\} \\ \nonumber && +\bm{X_{ct}\gamma}+\mu_t+\phi_c+\varepsilon_{ct}.
\end{eqnarray}
The outcome $Y_{ct}$ varies at the country $c$ and year $t$ level. The outcomes are initially the proportion of women in parliament (first stage compliance) and the natural logarithm of the maternal mortality ratio. We model a series of additional outcomes including intermediate outcomes (potential mechanisms), placebo outcomes and potential confounders. $Quota_{c}$ is 1 if a country ever adopted a quota, and this is interacted with a full set of leads and lags with respect to the year the quota was adopted. We include 10 lags and leads, the tenth term including all years greater than 10, and the first lead is omitted as the base category. We provide results varying this window down to 5 and 8 years. We include country and year fixed effects ($\phi_c$ and $\mu_t$), and cluster standard errors at the country level \citep{Bertrandetal2004}. 
    

   
\textbf{Parallel trends.} The $\beta^{lag}$ coefficients capture dynamic impacts and the $\beta^{lead}$ coefficients provide a partial test of the identifying assumption of no differential trends. This is only a partial test because, to estimate unbiased parameters we require parallel trends between treated and non-treated units \emph{in the absence of treatment}. Parallel pre-trends support this assumption but cannot be used to test what would have happened at the time of the reform had the reform not been implemented \citep{KahnLangLang2018}.  In view of concerns about the inclusion of unit-specific linear trends (Goodman-Bacon, 2019), we use the \citet{RambachanRoth2019} ``Honest DiD'' estimator that provides bounds on the post-quota coefficients under the scenario that any prevailing (even if imprecisely estimated) trends in the pre-treatment period between quota and non-quota countries are projected forward into the post-quota period (rather than assuming parallel trends going forward). We also use two estimators that make weaker parallel trends assumptions than our main approach. We implement the strategy of generating a synthetic control for each treated country (following \citeauthor{Abadieetal2010}, \citeyear{Abadieetal2010}; \citeauthor{Cavalloetal2013}, \citeyear{Cavalloetal2013}) by matching on pre-quota trends in maternal mortality, giving preference to countries within the same sub-region of the world (see Appendix \ref{App:Match}).  This synthetic control procedure chooses for each treated country its own synthetic control, and then pools pre- and post-treatment differences across countries to generate a single event-style plot.  We additionally use the recent synthetic difference-in-differences approach of \citet{arkhangelsky2021synthetic}, which allow us to generate a synthetic counterfactual to optimally align pre-treatment trends for \textit{each} quota year specific adoption group, resulting in a treatment vs.\ synthetic counterfactual comparison for each quota adoption period, as well as an aggregate ATT estimate.

\textbf{Staggered adoption.} A series of recent papers analyze the inference problem when treatment is staggered across units (countries) over time, creating multiple experiments. If there are heterogeneous treatment effects across countries or time, estimates obtained using the conventional difference in difference estimator may be biased.
We address this by using using the dynamic estimator of  \cite{dCDH2020}, which provides unbiased estimates. It uses groups whose treatment status is stable to infer the trends that would have affected switchers if their treatment had not changed.  Specifically, we report their `DID$_M$' estimate, which is based on a series of time-specific aggregated estimates.  
This estimator considers changes immediately surrounding quota adoption. It can be extended to estimate full dynamic impacts if rather than considering changes between $t$ and $t-1$, we consider changes between $t+k$ and $t-1$ for $k=1,2,\ldots,10$ years post reform, and similarly, placebo estimates can be presented showing changes between pre-reform periods
in which quota adopters have not yet been exposed.\footnote{The conventional event study pre-trends test \citep{Autor2003} has been shown to be invalid when treatment effects are heterogeneous \citep{abraham2018estimating} but placebo coefficients estimated following \cite{dCDH2020} are robust to this.} Inference is consistently based on a block bootstrap procedure, resampling over countries. Given the preferred properties of the \citet{dCDH2020} estimator in cases with heterogeneous impacts of treatment, we use this estimator as our main specification. Standard event studies described in equation \ref{eqn:wp-q} are provided as supplementary results.

\textbf{Sensitivity to estimator.} We supplement the dynamic models described above with single coefficient estimates of a two-way fixed effects (TWFE) specification in which the independent variable is defined as one for all years following the implementation of a quota for implementing countries, and zero before. It is set to zero for all countries that do not implement quotas in the sample period. The TWFE specification will tend to estimate an average of treatment effects that over-weights short-run effects and under-weights long-run effects \citep{BorusyakJaravel2017}. Since our event study estimates show that treatment effects increase with time since the quota event, in our setting the TWFE model will tend to produce conservative estimates. We provide a formal decomposition of the two-way FE estimator following \citet{GoodmanBacon2018}, quantifying how much identifying variation is drawn from a pure treatment versus control comparison and how much is drawn from variation in treatment timing. We additionally document how this single coefficient estimate compares to pooled coefficients from the event study, the aggregate dynamic estimates from \citeauthor{dCDH2020}'s ``$DID_M$'' estimator, and estimates using the  synthetic difference-in-differences approach of \citeauthor{arkhangelsky2021synthetic} which make comparisons in a more `local' way, giving more weight to control units which share more similar past trends with treated units, and more weight to time periods which are more similar to treated periods. Both the ``$DID_M$'' estimator and the synthetic DID estimator are robust to the biases inherent in standard two-way fixed effects models.\footnote{Nevertheless, the two-way fixed effects model provides a useful summary statistic. It is also helpful in assessing results for some intermediate outcomes (mechanisms) for which the data are relatively sparse, and when we use the \cite{RambachanRoth2019} and IV estimators to compute bounds.} 


\textbf{Additional robustness checks.} We directly investigate the threat to identification posed by the possibility that quota adoption responds to societal preferences becoming more pro-female, and that it is these underlying shifts in preferences that drive MMR decline. We similarly investigate democratization and a series of predictors of quota adoption discussed in the political science literature. We replicate the results for the staggered adoption across the Indian states of gender quotas in local area councils. This is useful not only because it is another setting but also because, in this case, quota implementation has been shown to have been quasi-random. The main estimates include the many (high and low income) countries that do not pass quotas in the sample period, as this expands the set of good comparisons available to identify trends \citep{BorusyakJaravel2017}. In different specification checks, we first drop the 51 (never treated) high income countries from the sample, and then restrict further to African countries, thus making the sample more homogeneous. The baseline estimates control only for country and year fixed effects. We display sensitivity of the estimates to adding a series of controls, $\bm{X_{ct}}$, including country-specific linear trends, region-year fixed effects, education and demographic characteristics of mothers, ethnic fractionalization and potentially endogenous intermediate outcomes including international development assistance for maternal health, GDP, public health expenditure, indicators of women's rights and gender equality, and political predictors of quota implementation. The potentially endogenous variables are consistently included as the baseline (pre-quota, in fact pre-1995) value of the variable multiplied by an indicator for post-quota years. This allows us to capture their direct impact on maternal mortality, also allowing that this impact changes after quota adoption.
We also show results restricting to a balanced panel and we address uncertainty in the MMR data by removing from the sample all modelled data, adjusting inference for the uncertainty, and by using the DHS to construct comparable survey data measures of MMR. 

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Results}
\label{scn:results}
\noindent\textbf{Main results.} 
The \citet{dCDH2020} estimates
are presented in Figure \ref{fig:DIDM}. They are obtained by aggregating estimates of outcome changes between adopters and non-adopters, comparing periods surrounding adoption. Panel (a) shows a discrete jump in women's parliamentary representation in the year after quotas are implemented. Panel (b) similarly shows a break in the coefficient series, with maternal mortality falling more rapidly after quota implementation. Panels (c) and (d) show that the patterns are broadly similar in models with no time-varying controls. The post-quota changes are persistent. Figure \ref{fig:robustness2alt} shows numerous specification checks, which we elaborate in the next section. The standard event study estimates (equation \ref{eqn:wp-q}) are appended as Figure \ref{fig:eventReserved} and they too provide broadly similarly estimates. 

Table \ref{tab:2wayFEcomp} demonstrates that our results are not sensitive to the choice of estimator. We show estimates from the two-way fixed effects single coefficient model, the weighted average of the dynamic coefficients from \citet{dCDH2020}, where the weights are the number of countries contributing to the coefficient for each year, and estimates from the synthetic difference-in-differences model. Following quota adoption we see an increase in the proportion of women in parliament of 5.7 to 6.6 percentage points which, relative to the baseline  average in 1985--1990 of 9\%, represents about a 64\% increase. We see a fall in the maternal mortality ratio of 7.2 to 12.7\%.\footnote{The median (mean) gender quota is 21\% (20\%). The estimated impacts of quotas on the proportion of women in parliament are smaller than the entire size of quotas. In quota implementing countries the pre-quota share of women in parliament was not always zero, the average was 7.9\%, rising to 20.9\% post-quota (median: 6.2 and 21.0\%). Taking all countries, the mean was 14.1\%, median 11.5\% (see Figures \ref{womparlPropAll} and \ref{womparlProp} for full distributions). See Figure \ref{quotacountry} for temporal variation by country. In Rwanda we see a jump in line with quota legislation but from a high baseline, while Djibouti shows a sharp jump from zero to quota attainment. The event studies of course show the unrestricted dynamic coefficients. We also show a series of estimates leaving out one country at a time.}

After quotas are legislated, the share of women in parliament can only change at the next election. For every country, we identified the years between quota legislation and the next election. The mode and median are zero years, the mean is 1.3.\footnote{In a few cases, it took time from quota passage until fulfillment. In Niger, for instance, the quota was in 2000 but the next election in 2004.} The event study plot shows a slight jump five years after the quota is passed, consistent with the subsequent election presenting a further opportunity to increase the share of women. Once women are in parliament, any changes they make will need to translate to the field and cumulate to have a discernible population-level impact on maternal mortality. Our finding that impacts of quota adoption on MMR increase with time since quota in Figures \ref{fig:DIDM} and \ref{fig:eventReserved} are consistent with this. Ten years out, MMR was 13\% lower in countries that passed quotas. 

The placebo coefficients are tightly centered on zero, showing no prevailing differential trends in MMR prior to quota adoption.\footnote{This is also the case when we use the long placebo indicated in \citet{dCDH2022}, see Appendix Figure \ref{fig:DIDMlongplacebo}.} 
We show in the next section that the bounds on the dynamic effects on MMR are consistent with our main findings when we relax the pre-trends assumption, and that our estimates hold with the synthetic control and synthetic DiD approaches. We also show that our findings hold in bounded IV estimates that allow that the exclusion restriction is violated and that the instrument is weak. We examine numerous potential predictors of quota adoption and show that they do not trend upward before the event of quota adoption, that they do not predict MMR and that our estimates are robust to controlling for them.

\noindent\textbf{Heterogeneity in impact.} Although international conventions suggested a target of 30\%, the mandated quota varied across countries, see Figure \ref{sizeQuotas} and Table \ref{tab:qclass}. Leveraging this variation, we find a clear ``dose-response'' (Figures \ref{fig:eventQuotaSizeDID} and \ref{fig:eventQuotaSize}). This increases our confidence that we identify impacts of quotas rather than an omitted variable, and it is also relevant to policy making. Estimates from the single-coefficient model (Table \ref{difdif-full}, panel B) indicate that quotas of less than 10\% increase the share of women in parliament by 2.8 percentage points, and generate a 0.6\% reduction in MMR, 
while quotas of 20--30\% increase the share of women in parliament by 6.8 percentage points and reduce MMR by 13.4\%. Baseline shares of women in parliament varied, as a result of which a given quota size had different impacts on the increase in the share of women in parliament. Baseline rates of MMR also varied considerably and we find larger declines where base rates were higher and there was more room for reduction. Estimates using terciles of baseline MMR are in Figure \ref{fig:eventQuotaSize}, panel B. We find a 7.7\% reduction in MMR in countries with a low baseline rate and a 15.9\% reduction in countries with a high baseline rate (Table \ref{difdif-full}, panel A).\footnote{We examine variation in baseline MMR and quota size jointly in Table \ref{difdif-full} panels A and B respectively, see columns 5 and 8.} 

\noindent\textbf{Effect sizes} Our estimates are of a similar order of magnitude to the historical introduction of occupational licensing requirements for U.S.\ midwives  \citep{anderson2020occupational}, the 10-year impact of  Brazil's Bolsa Familia conditional cash transfer program \citep{rasella2021long}, and the 4-year impact of Brazil's expansion of community-based primary care \citep{BhalotraRochaSoares2019}. In contrast, our estimates were larger than those from a large-scale program in India providing financial incentives for institutional birth (Janani Suraksha Yojana), which had no impact on maternal mortality \citep{powell2015financial}, and about 40\% as large as the impact of introducing antibiotics to treat high-risk, childbirth-related maternal infections in America in the late 1930s, when MMR was at a similar level to modern day Sub-Saharan Africa \citep{jayachandran2010modern}. 

Our IV estimates indicate that a 1 percentage point increase in women in parliament reduces MMR by 1.5 to 2\%. The elasticity, corresponding to a 1\% increase is 0.11 to 0.13.
We conducted a back of the envelope calculation using our estimates and rates of maternal mortality at the end of our sample period to project forward.  We estimate that the adoption of parliamentary gender quotas in all non-adopting countries could reduce rates of maternal mortality in Africa by 7.1\%, in Oceania by 1.6\%, in Asia by 1.3\%, in the Americas by 0.8\%, and in Europe by 0.1\%.


\noindent\textbf{Candidate list quotas} We used a similar approach to investigate impacts of candidate list quotas. These result in significant but smaller increases in women's share in parliament than reserved seat quotas. This is what we would expect since candidate quotas do not guarantee seats in parliament, also see \citet{PandeFord2012,BaguesEsteve2012}. We find no impact on MMR (Appendix Figures \ref{fig:DIDMCandidates}--\ref{fig:eventCandidates}). This seems plausible both because of their smaller impact on women's political representation, and because the countries implementing candidate list quotas during the study period (predominantly in Latin America) had achieved dramatic declines in MMR prior to quota implementation. This is backed by our estimates for reserved seat quotas showing quota impacts decreasing in baseline MMR, consistent with diminishing returns to policy intervention.

\noindent\textbf{Sub-national estimates for India.} Implementation of gender quotas in India was randomized at the village level, creating an opportunity for identification of their impact. However there are no MMR data at the village level. We leverage the fact that state level legislative reform implementing the village level gender quotas has been shown to be as good as randomly assigned \citep{Iyeretal2012}. Using the staggered adoption of quotas across India's states yields the estimates in Figure \ref{event-IndiaW}. Although less precise given the more limited variation available, these estimates, which rely on a source of identification that is distinct from our main cross-country analyses, confirm our main finding of quota-led declines in maternal mortality of 14.2\% (Table \ref{tab:India2wayFE}). This result is robust to leaving out one state at a time, see Figure \ref{LOOIndia}.

\section{Robustness Checks} 
In this section we summarize the main robustness checks. In Appendix \ref{app:Robustness}, we elaborate the checks discussed here and a few additional checks.

\label{sscn:robustExtend}
\noindent\textbf{Alternative estimators and pre-trends.} 
If standard tests of pretrends are underpowered, we might fail to capture the evolution of a relevant unobservable trend. To address this concern, we follow the ``Honest DiD'' procedure of \citet{RambachanRoth2019} and estimate upper and lower bounds on the dynamic effects. For both women in parliament (Figure \ref{honest-womparl}) and MMR (Figure \ref{honest-mmr}) these bounds are informative at 95\% or, in some cases in later periods, at 90\%.

Motivated by the concern that countries that do and do not adopt quotas may be different, we generate synthetic controls for each treated country. We aggregate estimates of treatment leads and lags across all treatment--synthetic control pairs, and implement an inference procedure based on clustered permutation to generate confidence intervals for these estimates, see Appendix \ref{App:Match}. The results in Figure \ref{fig:matchedControls} show no evidence of differential pre-trends, and a significant post-reform decline in MMR. We also re-estimate the models using the synthetic difference-in-differences procedure of \citet{arkhangelsky2021synthetic} and  observe, if anything,  slightly larger effects, see panel D, Table \ref{tab:2wayFEcomp}. 


Table \ref{tab:2wayFEcomp} shows that summary effect sizes from alternative estimators are similar in magnitude. This suggests that, in our setting, the potential bias in the single coefficient two-way FE model discussed in \citet{GoodmanBacon2018} and \citet{dCDH2020} is  small. To illuminate this we provide the \citet{GoodmanBacon2018} decomposition of the identifying variation into its treatment vs.\ pure control and differential timing components (Table \ref{tab:GBglobal}). The drop in MMR (of about 7\%) is similar when we compare early to late adopters (prior to adoption) to that obtained when comparing aggregate TWFE estimates of treated vs.\ never treated countries, albeit the weight attached to the latter is much greater (panel (b)).
Figure \ref{fig:GB} reveals estimates closely clustered around the average effect (indicated by the dashed red line), which suggests that the observed reduction in MMR is observed broadly. This is confirmed in Figures \ref{fig:DIDMLOO-MMR}--\ref{fig:LOO-WP} which show leave-one-out estimates.


To complement the reduced form estimates, we estimated LIML regressions of MMR on the share of women in parliament, instrumented with quota implementation.  The IV estimates in Table \ref{qIV} provide the scaled impact of women's parliamentary representation among compliers. They indicate that a 1 percentage point increase in women's share in parliament is associated with a 1.5 to 2.0\% decrease in MMR. If quotas were proxying an omitted variable the exclusion restriction would fail, but inference can still proceed on the premise of ``plausible exogeneity'', which delivers bounds on the IV estimates \citep{Conleyetal2012}. If we allow quota adoption to have a direct impact on MMR of up to $-$1\% over and above its impact on MMR via women in parliament, the bounds are informative, indicating a 0.01\% to 3.7\% reduction in MMR for a 1 percentage point increase in the women's share in parliament. We provide weak IV robust Anderson-Rubin confidence sets in Table \ref{qIV} and plot the implied rejection probabilities for the null across a range of null hypotheses in Figure \ref{fig:weakIV}, both of which suggest that the observed relationship between women in parliament and MMR is robust to corrections for weak instruments. 

\noindent\textbf{Selection into gender quota legislation.} The multiple approaches to investigating pre-trends discussed in the preceding section allay the key identification concerns. We nevertheless directly investigate the possibility that our results derive from social preferences evolving gradually to favor gender equality, with gender quota legislation being one manifestation of this. To measure social preferences and gender progressiveness in the policy environment, we pulled together data on 18 indicators of gender progressivity in the political, economic and civil domains including  indices of women's civil liberties, access to justice, economic rights, women's protests and the passage of abortion law (see the Data Appendix \ref{app:Data}). To examine the concern that the timing of quota adoption is a response to an upward drift in women's position, we use the same empirical strategy as for the main analysis, focusing on whether any of these 18 indicators shows an uptick prior to quota adoption. 
The estimates are in Figures \ref{fig:DIDMWomenRights2} and \ref{fig:DIDMWomenRights}.  The placebo coefficients (\citeauthor{dCDH2020}) allow us to reject a positive pre-trend for each of the 18 indicators. 

These findings are plausible. While the progression of gender equality in society is likely to eventually culminate in increased attention to women's reproductive health, this is likely to be a slow process. In contrast, we discover that giving women instrumental power to directly influence policy can effect sharp change. We will pursue evidence of this in the section on mechanisms, where we use the same identification strategy to test for impacts of discontinuous changes in women's political power on a series of intermediate outcomes that (a) can be influenced by policy and (b) are known to bring MMR down.

To investigate whether the estimated impacts of quota adoption might instead reflect  political changes \citep{Krook2009,BainesRubioMarin2005}, we follow the same approach as for women's rights, finding no pre-trend in seven measures of the political environment, see Figure \ref{fig:dCDHPoliSci}.


Although we find limited evidence that the 25 potential predictors, including women's rights and political variables, predict either quota adoption or MMR, we control in the main analysis for an index of the baseline (pre-quota, 1995) value of the indicators interacted with a dummy for post-quota years.
The estimated impact of quotas on MMR is not statistically different from the baseline estimate, see Figure \ref{fig:robustness2alt}, as well as full dynamic estimates with inference in Figures \ref{fig:DIDMReservedPred} and \ref{fig:eventReservedPred}. We also tested a series of placebo trend breaks before the true date of quota legislation in each country, and find no evidence of a pre-legislation break in trend in MMR (Table \ref{trendbreak}). 



\noindent\textbf{Endogenous changes in the composition of women giving birth.} 
If gender quotas lead to a shift in the composition of births such that women with lower baseline risks of maternal death are over-represented after quotas, then this could explain our finding of lower MMR. A compositional shift is not implausible given that we find an impact of quotas on fertility (Section \ref{scn:mechanisms}). We cannot investigate compositional change using aggregate country-year data on fertility, we need information on mother characteristics. To achieve this, we created a psuedo panel of births in the DHS data based on 10,837,442 births for 3,079,298 individual women from 82 countries surveyed in a total of 34 different years. We model quota impacts on birth rates of women of different education and age categories, and on the sex ratio (male/female) of births, a proxy for fetal and hence maternal health \citep{Waldron1983,low2000sex}. We find no significant shifts in composition by any of these measures, see Figures \ref{fig:DHScharacDIDM}--\ref{fig:DHS}.
Our estimates are robust to controlling for time-varying measures of the age and educational composition of mothers, see Figure \ref{fig:robustness2alt} and Figure \ref{fig:eventAlternativesQuota} panel d).


\noindent\textbf{Measurement of maternal mortality.} MMR varies considerably across countries, and proportional changes will exaggerate achievements in countries with lower baseline rates \citep{deaton2006global}. We find that the results hold when we replace the logarithm with the level of the MMR ratio (see Table \ref{difdif-full}).
The United Nations MMR data derive from vital statistics and demographic survey data, with gaps filled using modelled predictions \citep{Alkemaetal2015,Alkemaetal2016,WHO2015,Wilmothetal2012}. About 76\% of the country-year observations are original survey data points, and 24\% are imputed. Figure \ref{fig:robustness2alt} shows that removing countries for which all observations are imputed has no substantive impact on the findings. We directly account for this uncertainty, using a double-bootstrap procedure re-sampling over the provided uncertainty intervals to calculate the standard errors, see Table \ref{tab:resample}.
Allowing for correlation within country reduces the estimated uncertainty. For the \citet{dCDH2020} DID$_M$ estimator and synthetic DID, with a triangular resampling procedure, the main estimates are upheld.
We also use a measure of MMR which we derive from survey-based reports of sister deaths of DHS respondents, following the procedure detailed in \citet{BhalotraClarke2019}, albeit for fewer countries. See Figures \ref{DIDM-MMRDHS} and \ref{event-MMRDHS}). While less precisely estimated, the effects are larger, consistent with the DHS countries having higher MMR on average.

We considered whether the availability and quality of MMR data is endogenous. If women parliamentarians, motivated by a concern to reduce MMR, improve surveillance and tracking of MMR (``we measure what we treasure"), this will render our estimates conservative. However, this could go the other way if women politicians face more scrutiny of MMR and they react by pulling back on counting maternal deaths. Alternatively, if women politicians act to reduce the variance of (mean zero)
measurement error in MMR, this could make finding significant effects more likely. We acknowledge this problem without directly addressing it, but the fact that we identify mechanisms consistent with actual reductions in MMR suggests that our findings are less likely to be driven by changes in MMR measurement.



\noindent\textbf{Sensitivity to sample and clustering.} 
To assess sensitivity to changes in the composition of countries in the panel, we dropped the 7 countries that passed quotas after 2005 to create a balanced sample with the baseline window of 10 years pre and post-quota. The estimates are  unchanged (Figure \ref{fig:robustness2alt}). 
For some of the mechanisms variables analysed in the next section, the data are relatively sparse. We therefore re-estimate the main results on the common sample, see Figure \ref{fig:eventMechanismBAL}. The results are less precise in this smaller sample but the broad patterns remain. 
To address the potential concern that quota implementation was temporally correlated, we estimate event studies with two-way clustering \citep{Cameronetal2011} of standard errors by both country and by year, see Figures \ref{fig:eventAlternativesQuota}--\ref{fig:eventAlternativesQuotaWP}. While the confidence intervals are now wider, we still observe statistically significant effects. 




\section{Mechanisms}
\label{scn:mechanisms}

\noindent\textbf{Reproductive health coverage.} We investigate whether gender quotas influence the reproductive health services that have become conventional wisdom for policy \citep{WHO2014,Jamisonetal2013}, based on scientific consensus on their relevance to MMR reduction. Antenatal care is critical to identifying life threatening conditions such as pre-eclampsia and eclampsia early on, and having births attended by a skilled professional can reduce mortality from uterine bleeding and post-partum infection \citep{WHO2014,Jamisonetal2013}.
Contraceptive coverage may reduce fertility, and high fertility is a proximate cause of MMR \citep{girum2017correlates}.
Contraceptive coverage can also lower MMR without changing fertility by lengthening birth spacing or by substituting unsafe abortion \citep{MillerValente2016}. 

Figure \ref{fig:DIDMMechanism} panels (a)--(c) shows increased rates of coverage along these three dimensions of reproductive health in the years following quotas (and this also holds in the standard event study, see Figure \ref{fig:eventMechanism}). The single coefficient models (Table \ref{mDDmech}) show statistically significant increases in the share of coverage (in percentage points) of 5.8 in skilled birth attendance, 4.7 in prenatal care and (less precise estimates) of 1.7 in modern contraceptive use.\footnote{These results are robust to the alternative estimators discussed, except for contraceptive cover which shows an increase with \cite{dCDH2020} but not with \cite{RambachanRoth2019}.} Univariate descriptive associations of MMR with reproductive health coverage indicators on our analysis sample are in Table \ref{tab:univarMech}. A 1 percentage point increase in the share of attended births, prenatal care and access to contraception respectively is associated with declines in MMR of 4.4\%, 4.0\% and 6.3\%, magnitudes that make the identified impacts on MMR plausible.

\noindent\textbf{Abortion legislation, women's rights, and women's economic participation.} We investigate whether political power for women led to pro-female legislation being enacted, or to women being more likely to join the labor force. We find no evidence that quotas result in improvements in women's rights or economic participation-- so these are unlikely to be relevant  mechanisms. The only outcome of the eighteen we consider that responds to gender quotas is the political participation of women. This evidence is gathered from analysis of the indices of gender-related progressivity in Figure \ref{fig:DIDMWomenRights2} and Figure \ref{fig:DIDMWomenRights}. Specification checks based on standard event studies and \cite{RambachanRoth2019} are provided as Figures \ref{fig:eventWomenRights2}, \ref{fig:eventWomenRights}, \ref{fig:RRWomenRights2} and \ref{fig:RRWomenRights}. We also reproduce these figures conditioning upon country-specific linear trends, see Figures \ref{fig:DIDMMechanismtrend}--\ref{fig:DIDMWomenRightstrends}.\footnote{In Figure \ref{fig:DIDMWomenRights}, the only outcome of the set we investigate that responds to quota reform is women's labor market participation, which shows a decline. However, once we condition upon country-specific linear trends there is no impact. The results for the other outcomes are not sensitive to country-specific trends.}
In the discussion of identification, we pointed to the placebo coefficients of these estimates, and now we focus on the dynamic post-quota coefficients. 
One of the outcomes we investigate that merits particular attention is legalization of abortion, which directly related to women’s reproductive health because barriers to safe abortion lead to unsafe abortion, a major cause of MMR \citep{girum2017correlates}. Using data from \citet{Eliasetal2017}, we find no impact of quotas on abortion legislation, see Figures \ref{fig:DIDMWomenRights} and \ref{fig:eventWomenRights}. A possible reason is that there are strong religious barriers to abortion law which women leaders may not be able to surmount. 


\noindent\textbf{Education and fertility.} We investigated whether an increased presence of women in parliament modified female education and fertility, among demand side determinants of MMR. \cite{BhalotraClarke2013} provide quasi-experimental evidence that expanding female education brings MMR down, and many studies document a positive association of fertility with MMR \citep{girum2017correlates}. We find an increase of 0.5 years in the education of young women and a decline in the total fertility rate of 6--7\%, consistent with the observed expansion of contraceptive coverage and women's schooling.\footnote{Substantively similar results are observed in standard event-study estimates (Figure \ref{fig:eventMechanism}), in ``Honest DiD'' bounds (Figure \ref{fig:honestMechanism}) and when using a common sample across all mechanism variables (Figure \ref{fig:eventMechanismBAL}, though note that a small number of variables have sparse coverage, so estimation is noisier).}

On education, we study girls and boys aged 15--19 at the time of the reform, finding that attainment increases significantly more for girls than for boys (Figure \ref{fig:allEduc}, Table \ref{difdif-other}). This result is in line with evidence from India that quotas for women in local government led to an increase in girl's schooling, the suggested channel being  an increase in female aspirations \citep{beaman2009powerful}.  
The decline in fertility is accompanied by a (noisy) increase in birth spacing of 2 months (modelled using the DHS microdata), see Figure \ref{fig:DIDMMechanism}, consistent with the documented expansion of contraceptive coverage.

\noindent\textbf{Fertility -- parity and scale effects.} Since high fertility, defined as the number of children per woman, is associated with higher MMR risk per birth, a decline in fertility is a plausible mechanism for the observed decline in maternal death risk per birth (MMR). In addition, a decline in fertility will have a \emph{scale effect}, tending to reduce the number of maternal deaths at any level of risk per birth. Table \ref{tab:boeMDeaths} provides a back-of-the-envelope calculation of the number of maternal deaths averted on account of the impact of quotas on (a) MMR or risk per birth and (b) TFR or the number of births.\footnote{To do this we use data for the year before the imposition of quotas and sum across all quota countries to get baseline statistics of about 35 million births, and around 93 thousand maternal deaths in a year, corresponding to 266 deaths per 100,000 births. We apply the estimated declines of 8.2\% 
in maternal deaths per birth (Table \ref{tab:2wayFEcomp}) and 6.1\% in fertility (Table \ref{mDDmech}, column 4).} Our estimates are that from the baseline of 92,928 total deaths per year, post-quota deaths would fall to 84,843 (a fall of 8085 deaths) if only considering the MMR (per birth) channel, to 87,259 if only considering the scale effect of fertility (a fall of 5669 deaths), and to 79,668 (a fall of 13,260 deaths) if considering the total effect of gender quotas on the maternal death count, as mediated by both fertility and MMR decline. The scale effect (that is not captured in MMR decline) is roughly 43\% of the total change in the death count, and 64\% of the decline in deaths captured by MMR.

\noindent\textbf{Political change.} Earlier, we discussed political factors as potential predictors of quota legislation, focusing in Figure \ref{fig:dCDHPoliSci} on the placebo coefficients. We now study dynamic effects of quotas on the same range of political outcomes. We see that quota adoption is associated with a significant increase in the years that a regime is in power, and a corresponding decline in the probability of regime transition. In other words, after gender quota adoption, there is greater political stability, consistent with both being associated with constitutional reform.\footnote{\cite{Claytonetal2017}, for instance, argue that quotas are instituted with the intention of ensuring regime stability by bolstering support from women. She writes ``In Uganda, the ruling National Resistance Movement party implemented quotas in 1989 as part of a wider strategy to ensure regime stability and strengthen support among various social groups." Other authoritarian or semi-authoritarian regimes have similarly adopted gender quotas to bolster political support, see, e.g., \cite{panday2008representation} on Bangladesh; \cite{meena2004politics} on Tanzania; and \cite{longman2006rwanda} on Rwanda. In Eswatini, Kenya and Zimbabwe, quotas were brought about by the adoption of new constitutions, as was also the case in Nepal and Rwanda as they emerged from conflict. Our reading of the evidence, albeit for a few countries, is that violence and political instability create support for stable and inclusive governance and an appetite for constitutional reform.}
What is relevant for interpretation of our results is that our results hold conditional upon controls for regime stability.

We also investigated democratization, defined as in \cite{boix2013complete}. This is related to regime transition and is empirically relevant as developing countries (which dominate in quota adoption) often transition in and out of democracy. The preferred \citet{dCDH2020} DID$_M$ results are in Figure \ref{fig:democracyTests1} and event studies are provided in Figure \ref{fig:democracyTests2}. There is no evident tendency for quota adoption to increase democratization, or for democratization to lower MMR, albeit democratization has been shown to lead to lower infant mortality \citep{Kudamatsu2012}. The main estimates control for pre-quota democratization interacted with a post-quota trend, see Table \ref{tab:2wayFEcomp}, and Figure \ref{fig:DIDM}, panels (a) and (b). We perform a stricter test, controlling for a full set of lags and leads to democratic transitions. Our estimate of impacts of quotas on MMR is robust to this, see Figures \ref{fig:democracyTests1} and \ref{fig:democracyTests2}.

\section{Resources}
\label{scn:resources}
In this section we first address the substantive question of whether increasing the share of women in parliament benefits population health in general, or MMR in particular. This is related to the question of how any improvements are resourced. We therefore investigate if gender quota adoption is associated with increased resources, overall, and for health. We close the section with reference to existing studies, primarily single-country studies, with a view to illuminating how gender quotas can result in MMR reduction without a large increase in resources.

\noindent\textbf{Other population health outcomes.} In view of previous evidence that women leaders prioritize health \citep{miller2008women,bhalotra2014health,Bhalotraetal2019}, we investigate whether gender quotas led to generalized  improvements in population health or possibly detrimental impacts on other health outcomes, as would be the case if the observed improvements in MMR were achieved by allocating resources away from other population health priorities. To assess this we investigated adult male (and female) mortality, mortality from TB,\footnote{We chose this as it is high prevalence and gender-neutral. If anything, incidence and death rates from TB are higher among men than women. In 2017 close to 6 million adult men contracted TB and around 840,000 died from it. This compares with an estimated 3.2 million adult women who fell ill and almost half a million who died from TB (\href{https://www.who.int/tb/areas-of-work/population-groups/gender/en/}{WHO, September 2018}).} and infant mortality, which is widely regarded as a marker of population health in infectious disease environments such as those which characterize developing countries. DID$_M$ and event study plots are in Figures \ref{fig:DIDMGenderNeut}--\ref{fig:eventGenderNeut} and the TWFE coefficients in Table \ref{difdif-other}, columns 4--9.

Mortality among adult males, tuberculosis and infant mortality show no significant change following gender quota adoption, with quite tightly estimated zeros until at least 5 years post-quota.
There is some evidence of adult female mortality declining, but this is not statistically significant. This does not surprise us because any policies targeting causes of adult mortality among women other than MMR (TB, accidents, etc.) are not easily targeted to women. The reason that MMR is a conceptually clean outcome to study is that reproductive health services that address maternal mortality are by definition targeted at women.
Overall, the evidence points to (a) no  deterioration in the other population health outcomes studied, and thus no evidence of substitution, and (b) gender quotas being more effective at improving women's reproductive health and survival than in addressing other population health indicators. We suggest that both priorities and the potential to target women can explain why gender quotas have their largest impact on MMR.


\noindent\textbf{Resources and resource allocation.} We examine whether quota adoption led to an increase in available resources (Figure \ref{fig:DIDMMechanism}), which would constitute economic mechanisms. We find no evidence of an increase in GDP or in international development assistance for maternal health (DAH). We looked at DAH as this was increased after the year 2000 in response to MMR being included in the MDGs \citep{Dielemanetal2016}. It is of some interest to analyse the estimated coefficient on GDP. There is no impact of GDP on the share of women in parliament, but GDP has a significant direct impact on MMR. A 1\% increase in current GDP is associated with a decline in MMR of around 0.33\% (Table \ref{mechanismsRHS}).
A very crude back-of-the-envelope calculation assuming log-linearity (conditional on country and year FEs) suggests that to achieve the roughly 10\% reduction in MMR that we estimate as flowing from quota adoption, GDP would have to increase by nearly 30\%. This illustrates the force of women's political power. 


We tested for increases in state health expenditure. When this is measured as a share of GDP, the estimates are imprecise, see Figure \ref{DIDM-healtExp} for the \citeauthor{dCDH2020} estimates, and the standard event study plot in Figure \ref{event-healtExp}, also see Figures \ref{fig:healthSpendingDIDM} which fills in missing data. Alternative estimates normalizing on population rather than GDP (Figure \ref{fig:healthSpending}) show a significant increase (panels g, h). The single coefficient model indicates an increase of 0.89 percentage points or about 14\% following quota implementation (Table \ref{mDDmech}, column 7). As there is some weak evidence of pre-trends in the figures we account for this by producing ``Honest DiD'' bounds (Figure \ref{fig:honestMechanism}g) and now we see that the bounds are positive, with a lower bound of about 0.7 percentage points. 
Overall, there is some evidence of an increase in state health expenditure. However, our estimates are robust to controlling for GDP, DAH and health spending, see Figure \ref{fig:robustness2alt} and the corresponding DID$_M$ and  event study plots in Figures \ref{fig:eventAlternativesQuota}, \ref{fig:eventAlternativesQuotaWP}.\ref{fig:DIDMAlternativesQuota} and  \ref{fig:DIDMAlternativesQuotaWP}, 
Thus MMR reduction does not rely upon increasing public expenditure.




\noindent\textbf{Evidence on mechanisms from previous research.} In this section, we refer to a broader literature which illustrates the mechanisms by which leaders are able to move the outcomes they prioritize. These include legislation, advocacy, parliamentary (or council committee) debate, building consensus, information campaigns, role model effects (that raise the aspirations of younger women), committing resources to areas they prioritize, targeting existing resources, and improving coordination, management and efficiency of public service delivery (for example, by setting targets for civil servants). Women may act differently from men in these matters because they have different preferences, different information sets, are less corrupt, or have greater intrinsic motivation. This is potentially reinforced by women citizens feeling more able to raise their concerns with women representatives and raising their aspirations when exposed to female leaders.
%ASV: \footnote{Voting is often along party lines, and platforms too are often influenced by party leaders, so voting is not necessarily the most useful measures of political preferences.} 
 

On women changing the debate, recent analysis of text data illuminates the differentiated political preferences of women and the fact that they bring to parliament (or local council meetings), the issues that they care about \citep{Claytonetal2017,Lippmann2020,BaskaranHessami2019,Bhalotraetal2019}. Women in India's state legislatures have been shown to influence pro-female legislation \citep{ClotsFigueras2012}. Gender quotas in India have been associated with women citizens being more likely to be heard \citep{Iyeretal2012,Parthasarathy2019DeliberativeDI}. Using US data, \cite{GagliarducciPaserman2016} show that women are better at consensus-building, which is relevant if they want not only to generate debate but to achieve policy action. Building consensus for action on maternal mortality is likely to be harder than for action on child mortality. For example, one area where there is agreement between Republicans and Democrats in the US is early childhood investment.

There is some evidence that women bring resources to domains they prioritize, in particular, to health \citep{miller2008women,bhalotra2014health}. However resources may not be key for four reasons. First, policy action on the margins we identify in our analysis of mechanisms (namely, expanding the cadre of skilled prenatal and birth attendants and educating young women) is relatively low-cost because wages of the relevant personnel are low in developing countries. In particular, midwives, nurses and teachers can make a large difference to these outcomes-- see \cite{banke2020cost} on costs of extending prenatal care, and \cite{andrabi2020upping} on costs of extending schooling. Second, there is considerable scope to improve public services by limiting leakage of public funds on account of corruption. Women politicians in Brazil and India have been shown to be less corrupt than men, and less likely to distort policy to achieve electoral gains \citep{BrolloTroiano2016,Baskaranetal2018}. Third, there is evidence of substantial slack that good resource management can transform into productivity \citep{Bloometal2014,NBloometal2015}, including in the public health domain \citep{propper2010can}. In developing countries, public services including health services have been shown to suffer high rates of staff absenteeism \citep{WDR2004,DasHammer2014,Chaudhuryetal2006}. Correcting this sort of inefficiency does not require material resources as much as motivated governance. Women in politics appear to be more intrinsically motivated. 
If women have more information on MMR \citep{ashraf2020maternal}, they may also be better able to target resources. Fourth, our measures of reproductive health coverage are not purely supply-side measures, they also reflect uptake. Previous work suggests that low-cost outreach, information provision and education of women can improve uptake  \citep{miller2008women,Dupas2011,bhalotra2014health,Bhalotraetal2019,beaman2009powerful,CurrieMoretti2003}.

\section{Conclusion} 
\label{scn:conclusions}

Using longitudinal cross-country data across a period of 25 years that encompasses periods of dramatic but uneven decline in maternal mortality (and additionally using cross-state longitudinal variation within India), we provide new evidence that gender quotas that raise the share of women in parliament lead to substantial declines in maternal mortality. Current international strategy to address MMR is focused upon extending reproductive health coverage, but there is no recognition among policy makers of the political economy constraints on achieving this. We provide the first systematic analysis of the impacts of the recent wave of implementation of gender quotas across countries.\footnote{See, for instance, the review by \citet{PandeFord2012}, who discuss the cross-country implementation of quotas, but provide evidence based only on implementation of local government quotas in India.} Our findings suggest that gender quotas can be an effective policy tool for maternal mortality reduction.

Despite a wave of gender quota implementation, 130 countries in the world have none. There is thus substantial room for maneuver. While significant progress has been made, especially since 2000, preventable maternal mortality remains high, the lifetime risk of maternal mortality being 1 in 45 among women in low income countries. Our findings have implications for the recently launched Global Health 2035 report, and the ambitious Sustainable Development Goals. We show that SDG 3.1 targeting maternal mortality reduction is complementary to SDG 5.5 targeting an increase in women's political representation.

The broader evidence on the success of quotas is mixed \citep{coate1993will,besley2017gender,PandeFord2012,Niederle2016,van2018gender}.  However, our findings cohere with previous evidence that increasing the share of women politicians influences policy choices in favor of policies that align with the preferences of women \citep{chattopadhyay2004women,taylor2003women,Swers2005,ClotsFigueras2012,BaskaranHessami2019}. It also resonates with research showing that women politicians are more likely than men to invest in public health \citep{miller2008women,bhalotra2014health}, and that women voters prioritize public health while male voters prioritize low taxes \citep{Campbell2004}. In contrast to most public health outcomes, maternal mortality is unique to women and thus easy to overlook in a male-dominated parliament, but naturally targeted towards or ``assignable'' to women.


The overall decline in MMR of 8--12\% that we identify as flowing from quota implementation compares favorably with the 44\% global decline in MMR that occurred over the 25-year period, averaged over countries that did and did not implement quotas. Similarly, the 5--8 percentage point increase in birth attendance, the 4--8 percentage point increase in prenatal care, and the 1.7 percentage point increase in modern contraceptive use that we demonstrate flow from quota passage compare well with the 12, 13 and 9 percentage point increases (respectively) achieved through the recent 25 years. The decline in MMR of 44\% since 1990 fell well short of the MDG target decline of 75\% \citep{hogan2010maternal,kassebaum2014global}, and yet the new Sustainable Development Goals (SDGs) have set a higher target (of less than 70 per 100,000 live births by 2030). This is a clear flag that some policy innovation is needed -- and we suggest gender quotas. 







    
\renewcommand{\footnotesize}{\footnotesize}
\clearpage
\end{spacing}
\begin{spacing}{1}
\bibliographystyle{chicago}
\bibliography{politicsMMR}
\end{spacing}

\clearpage
\newgeometry{margin=0.5in}
\begin{spacing}{1.3}

\captionsetup[figure]{list=no}
\captionsetup[table]{list=no}
\section*{Figures and Tables}
\begin{figure}[htpb!]
  \caption{Trends in gender quotas, women in parliament and maternal mortality}
  \label{fig:RawReserved}
  \subfloat[Women in parliament and ln(maternal mortality ratio)\label{fig:rawMMR}]{%
    \includegraphics[scale=0.65]{./results/main/WomParlMMR.eps}%
  }
  \subfloat[Reserved seats and women in parliament\label{fig:rawWP}]{%
    \includegraphics[scale=0.65]{./results/main/reservedSeatsWP.eps}%
  }
  \floatfoot{\textsc{Notes:} Raw trends in number of countries with parliamentary gender quotas, the percentage of women in parliamentary seats and the log of the maternal mortality ratio. Data sources are provided in the Data Appendix. The sample is a global sample of 178 countries for which we have annual data through 1990--2015.} 
\end{figure}
    
    
\begin{figure}[htpp!]
    \subfloat[\% of women in parliament with time-varying controls]{%
    \includegraphics[scale=0.9]{./results/main/DIDM_womparl.eps}%
    }
    \subfloat[ln(maternal mortality ratio) with time-varying controls]{%
    \includegraphics[scale=0.9]{./results/main/DIDM_lnMMR.eps}
    }
    \vspace{-2mm}

    \subfloat[\% of women in parliament with no controls]{%
    \includegraphics[scale=0.9]{./results/main/DIDM_womparl_nocontrols.eps}%
    }
    \subfloat[ln(maternal mortality ratio) with no controls]{%
    \includegraphics[scale=0.9]{./results/main/DIDM_lnMMR_nocontrols.eps}%
    }
    \vspace{-2mm}
    \caption{Gender quota impacts: \citeauthor{dCDH2020}'s $DID_{M}$ estimator and placebos}
    \label{fig:DIDM}
    \floatfoot{Notes: Panels (a) to (d) present estimates of the impact of quotas on women in parliament and maternal mortality following \citet{dCDH2020}.  This consists of estimating aggregate impacts comparing all changers with non-changers surrounding the time of reform.  Plots present 10 placebo estimates comparing switchers and non switchers in pre-reform periods (when switchers had not yet changed), and dynamic effects showing impacts between $t+k$ and $t-1$ $\ \forall k\in\{0,1,\ldots,10\}$.  Panels (a)--(b) present models controlling for baseline resource and democracy controls interacted with a post quota dummy for women in parliament (panel (a)) and ln(MMR) (panel (b)).  Panels (c) and (d) present models with no controls.  Additional specifications are provided in Figure \ref{fig:robustness2alt} below. All inference is conducted using a block bootstrap, and average effects and standard errors are estimated by pooling all immediate and dynamic effects.}  
\end{figure}



\clearpage
\begin{landscape}
\begin{figure} 
  \caption{Robustness of \citet{dCDH2020} $DID_M$ estimates to alternative measures and specifications}
  \label{fig:robustness2alt}
  \subfloat[ln(MMR)]{%
    \includegraphics[scale=0.76]{./results/main/manyDIDMspecs_lnMMRt1.eps}%
  }
  \subfloat[\% Women in Parliament]{%
    \includegraphics[scale=0.76]{./results/main/manyDIDMspecs_womparl.eps}%
  }
  \vspace{-2mm}
  \floatfoot{Notes: Alternative plots graph $DID_M$ estimates surrounding the passage of quota reforms.  Coefficients indicated by a hollow black square correspond to baseline  \citet{dCDH2020} estimates without controls.  Alternative specifications are documented  as labelled in the graph legend.  Balanced panel refers to a sample consisting only of countries which adopted quotas prior to 2005 and as such exist in the entire range of quota post-treatment lags.  ``Removing modelled'' removes from the sample any countries based only on modelled maternal mortality data.  ``Empowerment Controls +'' controls for 25 variables measuring predictors of quotas indicated by the political science literature, and measures of empowerment and women's rights.  Controls are consistently specified using a baseline index based on z-scores of each variable (standardized such that higher values capture more positive movements of the underlying measure), interacted with post-quota adoption indicators.  Alternative specifications discussed, including point estimated as well as confidence intervals are documented in Appendix Figures \ref{fig:DIDMAlternativesQuota} and \ref{fig:DIDMAlternativesQuotaWP}.} 
\end{figure}
\end{landscape}


\begin{landscape}
  \begin{figure}[htpb!]
  \caption{Post-quota coefficients based on ``Honest DiD'': women in parliament and maternal mortality}
  \label{fig:HonestMain}
  \subfloat[Percent of women in parliament\label{honest-womparl}]{%
    \includegraphics[scale=0.68]{./results/main/EventHonest_womparl.eps}%
  }
  \subfloat[ln(maternal mortality ratio)\label{honest-mmr}]{%
    \includegraphics[scale=0.68]{./results/main/EventHonest_lnMMRt1.eps}%
  }
  \floatfoot{\textsc{Notes:} Each post-quota coefficient from event study specification \ref{eqn:wp-q} is documented, along with valid inference under \citet{RambachanRoth2019}'s ``Honest DiD'' methods.  These correspond to the specifications presented in Figure \ref{fig:eventReserved}, panels (a) and (b).  Here, in place of assuming parallel trends in quota and non-quota countries, valid 95\% confidence intervals are constructed under the assumption that post-quota trends in quota countries relative to non-adopters would have followed their prevailing path from the pre-quota period, permitting violations of standard parallel trend assumptions.} 
    \end{figure}
\end{landscape}


\begin{landscape}
  \begin{figure}[htpb!]
  \subfloat[Percent of women in parliament\label{match-womparl}]{%
    \includegraphics[scale=0.9]{./results/main/pooledMatchesInference_womparl.eps}%
  }
  \subfloat[ln(maternal mortality ratio)\label{match-lnMMR}]{%
    \includegraphics[scale=0.9]{./results/main/pooledMatchesInference_MMRt1.eps}%
  }
  \caption{Matched synthetic controls}
  \label{fig:matchedControls}
  \floatfoot{\textsc{Notes:} Refer to notes to Appendix \ref{App:Match}. Coefficients are estimated based on a pooled synthetic control approach where for each quota country a synthetic control is chosen based on leads of the variable of interest (up to period -3), over-weighting units which come from the same region as the country of interest.  Averages of each lag and lead are taken across all treatment--synthetic control matches.  Inference is conducted by permutation, where each permutation consists of randomly assigning the same distribution of quota reforms (blocked by countries to ensure identical treatment paths over time) but to non-reforming countries.}
  \end{figure}
\end{landscape}

  



\begin{landscape}
\begin{figure}[htpb!]
  \caption{Mechanisms: \citeauthor{dCDH2020} estimates for impacts of gender quotas on intermediate outcomes}
  \label{fig:DIDMMechanism}
  \subfloat[Antenatal care\label{DIDM-antenatal}]{%
    \includegraphics[scale=0.72]{./results/main/DIDM_antenatal_ipol.eps}%
  }
  \subfloat[Attended births\label{DIDM-birthattend}]{%
    \includegraphics[scale=0.72]{./results/main/DIDM_birthAttend_ipol.eps}%
  }
  \subfloat[Modern contraceptive cover\label{DIDM-modernCont}]{%
    \includegraphics[scale=0.72]{./results/main/DIDM_contracepMod.eps}
  }
  \vspace{-4mm}
  
  \subfloat[ln(fertility)\label{DIDM-BirthRate}]{%
    \includegraphics[scale=0.72]{./results/main/DIDM_lnFert.eps}%
  }
  \subfloat[Teen pregnancy\label{DIDM-BirthRateTeen}]{%
    \includegraphics[scale=0.72]{./results/main/DIDM_teenPreg.eps}%
  }
  \subfloat[Birth spacing\label{DIDM-birthSpace}]{%
\includegraphics[scale=0.72]{./results/main/DIDM_birthSpacePost.eps}
  }
  \vspace{-4mm}
  
  \subfloat[Health expenditure as a \% of GDP\label{DIDM-healtExp}]{%
    \includegraphics[scale=0.72]{./results/main/DIDM_healthExpt1.eps}%
  }
  \subfloat[Development assistance for maternal health\label{DIDM-DAH}]{%
    \includegraphics[scale=0.72]{./results/main/DIDM_p_mh_dah_17.eps}
  }
  \subfloat[ln(GDP per capital)\label{DIDM-GDPpc}]{%
    \includegraphics[scale=0.72]{./results/main/DIDM_lnGDP.eps}%
  }
  \vspace{-8mm}
  \floatfoot{\textsc{Notes:} \citet{dCDH2020} $DID_M$ estimates of intermediate outcomes as a function of the passage of gender quotas are presented. Antenatal coverage and birth attendance refer to the percentage of coverage, are accessed from the World Bank databank, and are only available for a sub-sample of years for each country (an unbalanced panel from 1990--2015). 
  Fertility rates refer to the expected births per women (the total fertility rate) and are recorded as World Bank databank indicator SP.DYN.TFRT.IN.  Birth spacing is calculated as the average time to subsequent births for all women giving birth in a country and year based on DHS fertility rosters. Health expenditure is expressed as a percent of GDP, and is accessed from the  World Health Organization's National Health Accounts (NHA) data series. Proportion of development assistance for health that goes towards maternal health is provided by the Institute for Health Metrics and Evaluation (IHME) Development Assistance for Health Database. The log of GDP per capita is PPP adjusted and measured in 2011 international dollars.  All other details follow those of estimates presented in Figure \ref{fig:DIDM}.} 
\end{figure}
\end{landscape}

  
\begin{landscape}
  \begin{figure}[htpb!]
    \centering
    \subfloat[Quotas and Democracy]{
    \includegraphics[scale=0.82]{./results/main/DIDM_Democracy.eps}
    }%
    \subfloat[Democracy and Maternal Mortality]{%
    \includegraphics[scale=0.82]{./results/main/DemocracyDIDM_lnMMRt1.eps}
    }%  
    \subfloat[Quotas and MMR, conditional on Democracy]{
    \includegraphics[scale=0.82]{./results/main/DIDM_lnMMRt1_democracyCOND.eps}
    }
    \caption{Quotas, Democratic Transitions and Maternal Mortality Declines -- DID$_M$ Estimates}
    \label{fig:democracyTests1}
    \floatfoot{\textsc{Notes:} Here we consider the potential role of democratization in observed impacts of quota adoption on maternal mortality. Left-hand panel estimates the impact of quota adoption on whether or not a country is classified as a democracy.  Center panel estimates the impact of transition to a democracy (rather than quota adoption) on rates of maternal mortality.  Right-hand panel reports the impacts of quota adoption on maternal mortality when additionally including as controls full lags and leads to the adoption of democracy in cases where countries are classified as moving from non-democratic to democratic.  All other estimation details follow those in Figure \ref{fig:DIDM}.}
    \end{figure}
\end{landscape}



\begin{landscape}    
\begin{figure}[htpb!]
  \caption{Gender quotas and schooling (15--19 year-olds)}
  \label{fig:allEduc}
    \subfloat[Women (DID$_M$)]{%
    \includegraphics[scale=0.75]{./results/main/DIDM_yearsSchool1519_ipol.eps}%
  }\hspace{8mm}
  \subfloat[Men (DID$_M$)]{%
    \includegraphics[scale=0.75]{./results/main/DIDM_yearsSchool1519M_ipol.eps}%
  }\hspace{8mm}
  \subfloat[Women/Men (DID$_M$)]{%
    \includegraphics[scale=0.75]{./results/main/DIDM_schoolingRatio.eps}%
  }

   \hspace{-8mm}\subfloat[Women (Honest DiD)]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_yearsSchool1519_ipol.eps}%
  }\hspace{1.3cm}
  \subfloat[Men (Honest DiD)]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_yearsSchool1519M_ipol.eps}%
  }\hspace{1.3cm}
  \subfloat[Women/Men (Honest DiD)]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_schoolingRatio.eps}%
  }    
  \vspace{-8mm}
  \floatfoot{\textsc{Notes:} Impacts of reserved seat quotas on female education, male education, and the ratio of female to male education are presented.  Panels (a)-(c) present estimates using \citeauthor{dCDH2020}'s $DID_M$ estimator, and panels (d)-(f) present  estimates based on \citeauthor{RambachanRoth2019}'s Honest DiD estimator. Education measures the average years of schooling for 15--19 year-olds. }
\end{figure}
\end{landscape}

\begin{figure}[htpb!]
    \caption{Mechanisms: Post-quota coefficients based on ``honest DiD''}
    \label{fig:honestMechanism}
    \subfloat[Antenatal care\label{honest-antenatal}\vspace{-4mm}]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_antenatal_ipol.eps}%
    }
    \subfloat[Attended births\label{honest-birthattend}\vspace{-4mm}]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_birthAttend_ipol.eps}%
    }
    \subfloat[Modern contraceptive cover\label{honest-modernCont}\vspace{-4mm}]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_contracepMod.eps}
    }
    \vspace{-4mm}
    
    \subfloat[ln(fertility)\label{honest-BirthRate}\vspace{-4mm}]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_lnFert.eps}%
    }
    \subfloat[Teen pregnancy\label{honest-teenPreg}\vspace{-4mm}]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_teenPreg.eps}%
    }
    \subfloat[Birth spacing\label{honest-GirlsEduc}\vspace{-4mm}]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_birthSpacePost.eps}
    }
    \vspace{-4mm}
      
    \subfloat[Health expenditure as a \% of GDP\label{honest-healtExp}\vspace{-4mm}]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_healthExpt1.eps}%
    }
    \subfloat[Dev.\ assistance for maternal health\label{honest-DAH}\vspace{-4mm}]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_p_mh_dah_17.eps}
    }
    \subfloat[ln(GDP per capital)\label{honest-GDPpc}\vspace{-4mm}]{%
    \includegraphics[scale=0.35]{./results/main/EventHonest_lnGDP.eps}%
    }
    
    \floatfoot{\textsc{Notes:} Each post-quota coefficient from event study \ref{eqn:wp-q} is documented, along with valid inference under \citet{RambachanRoth2019}'s ``Honest DiD'' methods.  Here, in place of assuming parallel trends in quota and non-quota countries, valid 95\% confidence intervals are constructed under the assumption that post-quota trends in quota countries relative to non-adopters would have followed their prevailing path from the pre-quota period, permitting violations of standard parallel trend assumptions.}
    \end{figure}
    



\clearpage
%\section*{Tables}

\begin{table}[htpb]
\centering
\caption{Two-way FE, \citeauthor{dCDH2020}, pooled event studies, and synthetic DID estimates}
\label{tab:2wayFEcomp}
\scalebox{0.95}{
\begin{tabular}{lcccccc} \toprule
& (1)     & (2)     & (3)     & (4)     & (5)     & (6)     \\ \cmidrule(r){2-7}
& \multicolumn{6}{c}{Outcome: ln(MMR)} \\ \midrule
%\input{./results/main/ModelComp_lnMMRt1.tex}
\multicolumn{5}{l}{\textbf{Method A: Two-way FE Model}}\\ 
Reserved Seats&-0.082&-0.156*&-0.075&-0.106*&-0.071&-0.246*\\
    &(0.051)&(0.090)&(0.056)&(0.056)&(0.055)&(0.130)\\
\multicolumn{5}{l}{\textbf{Method B: $ DID\_M $ Estimates}}\\ 
Reserved Seats&-0.072*&-0.074*&-0.072*&-0.074*&-0.080*&-0.082\\
    &(0.043)&(0.043)&(0.043)&(0.043)&(0.047)&(0.050)\\
\multicolumn{5}{l}{\textbf{Method C: Pooled Event Study}}\\ 
Reserved Seats&-0.079**&-0.154&-0.076*&-0.106*&-0.058&-0.232\\
    &(0.039)&(0.100)&(0.042)&(0.058)&(0.045)&(0.166)\\
\multicolumn{5}{l}{\textbf{Method D: Synthetic DID}}\\ 
Reserved Seats&-0.127*&-0.116*&-0.129*&-0.103&-0.128&-0.099\\
    &(0.067)&(0.069)&(0.072)&(0.064)&(0.080)&(0.062)\\
\\ Negative Weights & -0.005 & -0.143 & -0.019 & -0.006 & -0.012 & -0.445 \\ 
   Observations & 4335  & 4241  & 4335  & 4241  & 4335  & 4241  \\ 
\midrule
& \multicolumn{6}{c}{Outcome: Women in Parliament} \\ 
\midrule
%\input{./results/main/ModelComp_womparl.tex}
\multicolumn{5}{l}{\textbf{Method A: Two-way FE Model}}\\ 
Reserved Seats&5.793***&6.333&6.071**&6.077**&6.038***&9.207\\
    &(2.167)&(4.521)&(2.478)&(2.645)&(2.145)&(6.266)\\
\multicolumn{5}{l}{\textbf{Method B: $ DID\_M $ Estimates}}\\ 
Reserved Seats&5.678**&5.674***&5.678**&5.674***&5.167**&5.128***\\
    &(2.222)&(1.880)&(2.222)&(1.880)&(2.154)&(1.872)\\
\multicolumn{5}{l}{\textbf{Method C: Pooled Event Study}}\\ 
Reserved Seats&6.622***&7.179&6.940***&7.079**&6.242***&9.668\\
    &(1.862)&(5.082)&(2.015)&(3.314)&(1.891)&(8.341)\\
\multicolumn{5}{l}{\textbf{Method D: Synthetic DID}}\\ 
Reserved Seats&8.281***&7.523***&8.361**&7.950**&7.661***&7.014**\\
    &(2.611)&(2.344)&(3.597)&(3.246)&(2.552)&(2.724)\\
\\ Negative Weights & -0.005 & -0.143 & -0.019 & -0.006 & -0.012 & -0.445 \\ 
   Observations & 4335  & 4241  & 4335  & 4241  & 4335  & 4241  \\ 
\midrule
\textbf{\small{Controls (baseline)}}:&&&&&&\\
\small{Empowerment \& Predictors} && \small{Y} & & & & \small{Y} \\ 
\small{Democracy}   && & \small{Y} & & & \small{Y} \\ 
\small{Resources}   && & & \small{Y} & & \small{Y} \\
\small{Region$\times$year FE}    && & & & \small{Y} & \small{Y} \\
\bottomrule
\multicolumn{7}{p{15.2cm}}{\footnotesize Notes: Each cell presents results from a separate individual reduced form estimate varying estimation procedures (in rows) and included controls (in columns).  The top panel considers the impact of reserved seats on maternal mortality, while the bottom panel considers the impact of reserved seats on women in parliament. Two-way FE models refers to linear regression controlling for country and year fixed effects.  $DID_M$ refers to pooled estimates based on \citeauthor{dCDH2020}.  Pooled event study estimates impacts pooling all post-event study coefficients (from lag 0 to lag 10+). Synthetic DID implements \citet{arkhangelsky2021synthetic}.  In each case, standard errors are estimated clustering by country or using a block bootstrap by country.  Controls are consistently generated using baseline (pre-1995 measures) interacted with a post-quota dummy.  At the base of each panel we present the magnitude of the negative weights attached to the two-way FE estimate, following \citet{dCDH2020}.  * p$<$0.10; ** p$<$0.05; *** p$<$0.01.}
\end{tabular}}
\end{table}






    
\clearpage
\appendix

\setcounter{page}{1}
\renewcommand{\thepage}{A\arabic{page}}
\begin{spacing}{1.3}
\label{app:figtab}
\begin{center}
{\Large Online Appendix -- Not for Print}\\
\textbf{Maternal Mortality and Women's Political Participation} \\
Sonia Bhalotra, Damian Clarke, Joseph Gomes, Atheendar Venkataramani
\end{center}
\end{spacing}

%\listoffigures
%\listoftables

\setcounter{table}{0}
\renewcommand{\thetable}{B\arabic{table}}
\setcounter{figure}{0}
\renewcommand{\thefigure}{B\arabic{figure}}




%------------------------------------------------------------------------------- 


\section*{Appendix Figures and Tables}



\renewcommand{\listfigurename}{List of Figures (Appendix and Online Appendix)}
\renewcommand{\listtablename}{List of Tables (Appendix and Online Appendix)}
\end{spacing}
\listoffigures
\listoftables
    
\begin{spacing}{1.3}    
\captionsetup[figure]{list=yes}
\captionsetup[table]{list=yes}
\setcounter{table}{0}
\renewcommand{\thetable}{A\arabic{table}}
\setcounter{figure}{0}
\renewcommand{\thefigure}{A\arabic{figure}}



\clearpage
\section{Appendix Figures}
\begin{figure}[htpp!]
  \subfloat[Female life expectancy\label{FLE_LGDP}]{%
    \includegraphics[scale=0.45]{./results/appendix/summary/lifeExpF_lnGDP}%
  }
  \subfloat[Male life expectancy\label{MLE_LGDP}]{%
    \includegraphics[scale=0.45]{./results/appendix/summary/lifeExpM_lnGDP}%
  }
	  \subfloat[Female/male life expectancy ratio\label{LEratio_LGDP}]{%
    \includegraphics[scale=0.45]{./results/appendix/summary/lifeExpR_lnGDP}%
  }	
  \caption{Life expectancy in 2010}
  \label{fig:LELnGDP}
  \floatfoot{\textsc{Notes:} Life expectancy at birth and PPP-adjusted GDP per capita data are collated by the World Bank Data Bank.  These indicators are SP.DYN.LE00.FE.IN (Female life expectancy) SP.DYN.LE00.MA.IN (Male life expectancy) and NY.GDP.PCAP.PP.KD (PPP adjusted GDP per capita).  The life expectancy ratio is calculated as female life expectancy divided by male life expectancy for each country. Lowess fits are overlaid on scatter plots, using a bandwidth of 0.8 for local linear smoothing.}
\end{figure}






    

    
%\clearpage
%\section*{Appendix Figures}
\begin{figure}[htpb!]
  \begin{center}
    \caption{Quota coverage: 1990--2015}
    \label{quotacover}
    \includegraphics[scale=0.7]{./results/auxiliary/coverage-both.eps}
  \end{center}
  \floatfoot{\textsc{Notes:} Geographic distribution of countries implementing a quota for reserved seats in parliament and candidate list quotas. Data is compiled from \citet{Dahlerup2005} and updated with information for recent years from the online quotaproject.org database developed and maintained by the International Institute for Democracy and Electoral Assistance (IDEA), the Inter-Parliamentary Union, and Stockholm University. This database was consulted on 19th of July, 2016.}
\end{figure}


\begin{figure}[htpb!]
  \begin{center}
    \caption{Maternal mortality ratio: 1990--2015}
    \label{MMRworld}
    \includegraphics[scale=0.7]{./results/auxiliary/MMRworld-eps-converted-to.pdf}
  \end{center}
  \floatfoot{\textsc{Notes:} Average rates by country
    for the period 1990--2015.  Values are calculated as deaths per 100,000 live births, and are provided by the MMEIG, an inter-agency project of the WHO, UNICEF, UNFPA, World Bank Group, and the United Nations Population Division.}
\end{figure}
   



\begin{figure}[htpb!]
  \begin{center}
    \caption{Reserved seat quota timing: 1990--2012}
    \label{quotatiming}
    \includegraphics[scale=0.7]{./results/auxiliary/quotasStackedRes-eps-converted-to.pdf}
  \end{center}
  \floatfoot{\textsc{Notes:} Timing of the implementation of reserved seats is documented by geographic region. Additional notes related to quota adoption are provided in Figure \ref{quotacover}.}
\end{figure}
    
\begin{figure}[htpb!]
  \begin{center}
    \caption{Reserved seat quota sizes}
    \label{sizeQuotas}
    \includegraphics[scale=1]{./results/appendix/summary/quotaSizes.eps}
  \end{center}
  \vspace{-6mm}
  \floatfoot{\textsc{Notes:} This histogram describes the quota
    size for each country which adopts a reserved seat quota.  Each country (quota) is included as a single observation.}
\end{figure}

\begin{figure}[ht!]
  \caption{Gender quota impacts on women in parliament and maternal mortality}
  \label{fig:eventReserved}
  \subfloat[Percent of women in parliament with time-varying controls\label{event-womparl}]{%
    \includegraphics[scale=0.52]{./results/appendix/alternative/origcontrols_event_womparl_10.pdf}%
      }
      \subfloat[ln(maternal mortality ratio) with time-varying controls\label{event-mmr}]{%
        \includegraphics[scale=0.52]{./results/appendix/alternative/origcontrols_event_lnMMRt1_10.pdf}%
      }

      \subfloat[Percent of women in parliament with no controls\label{event-womparl}]{%
        \includegraphics[scale=0.52]{./results/appendix/event/event_womparl_10.pdf}%
      }
      \subfloat[ln(maternal mortality ratio) with no controls\label{event-mmr}]{%
        \includegraphics[scale=0.52]{./results/appendix/event/event_lnMMRt1_10.pdf}%
      }
      \floatfoot{\textsc{Notes:} Point estimates of the lag and lead terms in the event study specification described in equation \ref{eqn:wp-q} are presented, along with their 95\% confidence intervals. Estimates are conditional on country and year fixed effects.  Panels (a) and (b) additionally control for the log of GDP per capita and an indicator of whether the country is a democracy. Time periods greater than 10 years from the reform date are displayed as a single ``10 +'' indicator. Standard errors are clustered by country. The omitted base category is taken as 1 year prior to the reform, indicated by the solid vertical line.} 
    \end{figure}


\begin{figure}[htpb!]
  \begin{center}
  \caption{Proportion of women in parliament}
  \subfloat[All Countries\label{womparlPropAll}]{%
    \includegraphics[scale=0.65]{./results/appendix/summary/quotasWomenParliamentAll.eps}%
  }
  \subfloat[Reserved Seat Countries Only\label{womparlProp}]{%
    \includegraphics[scale=0.65]{./results/appendix/summary/quotasWomenParliament.eps}%
  }
  \end{center}
  \vspace{-6mm}
  \floatfoot{\textsc{Notes:} Density plots describe the proportion of women in parliament in all countries and years under study (figure (a)), and the proportion of women in parliament in countries which at some point adopt a reserved seat quota (figure (b)).  Plots are based on each country by year observation in the women in parliament data.}
\end{figure}
    



\begin{landscape}
  \begin{figure}[htpb!]
    \begin{center}
    \caption{Country-specific changes in women in parliament after reserved seat quotas}
    \label{quotacountry}
    \includegraphics[scale=1.1]{./results/appendix/summary/ReservedCountries.eps}
    \end{center}
    \vspace{-8mm}
    \floatfoot{\textsc{Notes:} In each panel, the vertical lines display the recorded date of the passage of a reserved seat quota for women in the national parliament, and the plots show the evolution of the percentage of women in parliament. Each figure shares a common y-axis for ease of comparison.}
  \end{figure}
\end{landscape}


\begin{landscape}
  \begin{figure}[htpb!]
    \textbf{Panel A: By Quota Size} \vspace{4mm} \\ 
    \centering
    \subfloat[Quota Size: 0-10]{%
    \includegraphics[scale=0.8]{./results/appendix/heterogeneity/DIDM_lnMMRt1_010.eps}
    }%  
    \subfloat[Quota Size: 10-20]{
    \includegraphics[scale=0.8]{./results/appendix/heterogeneity/DIDM_lnMMRt1_1020.eps}
    }%
    \subfloat[Quota Size: 20-30]{
    \includegraphics[scale=0.8]{./results/appendix/heterogeneity/DIDM_lnMMRt1_2030.eps}
    }

    \vspace{4mm} \textbf{Panel B: By Baseline Rates of Maternal Mortality} \vspace{4mm} \\
    \subfloat[``Low'' baseline Rates (MMR $<$400)]{%
    \includegraphics[scale=0.8]{./results/appendix/heterogeneity/DIDM_lnMMRt1_Low.eps}
    }      
    \subfloat[``Medium'' baseline Rates (MMR [400,800))]{
    \includegraphics[scale=0.8]{./results/appendix/heterogeneity/DIDM_lnMMRt1_Mid.eps}
    } 
    \subfloat[``High'' baseline Rates (MMR $\geq$800)]{
    \includegraphics[scale=0.8]{./results/appendix/heterogeneity/DIDM_lnMMRt1_High.eps}
    }
    \caption{Impact of quotas on MMR by size of gender quota (DID$_M$ estimates)}
    \label{fig:eventQuotaSizeDID}
    \floatfoot{\textsc{Notes:} \citet{dCDH2020} $DID_M$ estimates are presented separately for countries with a different proportion of seats reserved in their quota (Panel A) or with different rates of maternal mortality at baseline (Panel B).  These are presented in three bins.  Bins approximately separate quota countries into three evenly sized groups. Baseline rates of maternal mortality are calculated as average values in countries prior to the year 2000.  Each set of coefficients is estimated as separate $DID_M$ models, removing all quota countries which are not classified as part of that group, and using all non-quota countries as untreated controls.  Remaining details follow Figure \ref{fig:DIDM}.}
    \end{figure}
\end{landscape}

    
\begin{landscape}
  \begin{figure}[htpb!]
    \textbf{Panel A: By Quota Size} \vspace{4mm} \\ 
    \centering
    \subfloat[Quota Size: 0-10]{%
    \includegraphics[scale=0.45]{./results/appendix/heterogeneity/event_lnMMRt1_010.pdf}
    }%  
    \subfloat[Quota Size: 10-20]{
    \includegraphics[scale=0.45]{./results/appendix/heterogeneity/event_lnMMRt1_1020.pdf}
    }%
    \subfloat[Quota Size: 20-30]{
    \includegraphics[scale=0.45]{./results/appendix/heterogeneity/event_lnMMRt1_2030.pdf}
    }

    \vspace{4mm} \textbf{Panel B: By Baseline Rates of Maternal Mortality} \vspace{4mm} \\
    \subfloat[``Low'' baseline Rates (MMR $<$400)]{%
    \includegraphics[scale=0.45]{./results/appendix/heterogeneity/event_lnMMRt1_Low.pdf}
    }      
    \subfloat[``Medium'' baseline Rates (MMR [400,800))]{
    \includegraphics[scale=0.45]{./results/appendix/heterogeneity/event_lnMMRt1_Mid.pdf}
    } 
    \subfloat[``High'' baseline Rates (MMR $\geq$800)]{
    \includegraphics[scale=0.45]{./results/appendix/heterogeneity/event_lnMMRt1_High.pdf}
    }
    \caption{Impact of quotas on MMR by size of gender quota (event study models)}
    \label{fig:eventQuotaSize}
    \floatfoot{\textsc{Notes:} Event study estimates are presented separately for countries with a different proportion of seats reserved in their quota (Panel A) or with different rates of maternal mortality at baseline (Panel B).  These are presented in three bins.  Bins approximately separate quota countries into three evenly sized groups. Baseline rates of maternal mortality are calculated as average values in countries prior to the year 2000.  Each set of coefficients is estimated in a single event study (one in Panel A and one in Panel B), and so are conditional on all other groups in the panel.  In each case, the baseline group consists of countries which do not implement a quota.  Remaining details follow Figure \ref{fig:eventReserved}.}
    \end{figure}
\end{landscape}

   
\begin{figure}[htpb!]
    \subfloat[\% of women in parliament]{%
      \includegraphics[scale=1]{./results/appendix/DIDM_womparl_LONGPLACEBO.eps}%
    }
    \subfloat[ln(maternal mortality ratio)]{%
      \includegraphics[scale=1]{./results/appendix/DIDM_lnMMRt1_LONGPLACEBO.eps}%
    }
\caption{\citet{dCDH2022}'s DID$_M$ estimator with long placebos}
\label{fig:DIDMlongplacebo}
\floatfoot{\textsc{Notes:} Results replicate those in Figure \ref{fig:DIDM}, however now using `long placebos' described in \citet{dCDH2022} which consider movements in pre-periods consistently coared with period -1, rather than short placebos based on movements of one period (eg from -5 to -4, or -3 to -2) during the pre-treatment period. Post-treatment estimators are identical in both cases.}
\end{figure}






\begin{figure}[htpp!]
\subfloat[Women in parliament]{%
\includegraphics[scale=1]{./results/appendix/event/Cand_DIDM_womparl.eps}%
}
\subfloat[ln(MMR)]{%
\includegraphics[scale=1]{./results/appendix/event/Cand_DIDM_lnMMRt1.eps}%
}
\caption{Candidate list quotas: $DID_M$ estimates for women in parliament and maternal mortality}
\label{fig:DIDMCandidates}
\floatfoot{Notes: Identical specifications are estimated as in Figure \ref{fig:DIDM} panels (a) and (b), however now with countries passing candidate list quotas. Average post-quota estimates and their block bootstrapped standard errors are provided in text on the plot.  Countries implementing candidate list quotas in the period under study are Albania, Angola, Argentina, Armenia, Belgium, Bolivia, Bosnia and Herzegovina, Brazil, Burkina Faso, Costa Rica, Croatia, Dominican Republic, Ecuador, El Salvador, France, Greece, Guinea, Guyana, Honduras, Indonesia, Ireland, South Korea, Kyrgyz Republic, Lesotho, Macedonia, Mauritania, Mexico, Mongolia, Montenegro, Nepal, Nicaragua, Panama, Paraguay, Peru, Poland, Portugal, Senegal, Serbia, Slovenia, Spain, Tunisia and Uruguay.}
\end{figure}



\begin{figure}[htpp!]
\subfloat[Women in parliament]{%
\includegraphics[scale=0.55]{./results/appendix/event/Candevent_womparl_10.pdf}%
}
\subfloat[ln(MMR)]{%
\includegraphics[scale=0.55]{./results/appendix/event/Candevent_lnMMRt1_10.pdf}%
}
\caption{Candidate list quotas: event studies for women in parliament and maternal mortality}
\label{fig:eventCandidates}
\floatfoot{Notes: Identical specifications are estimated as in Figure \ref{fig:eventReserved}, however now with countries passing candidate list quotas. Difference-in-difference estimates from a single-coefficient model suggest an increase in 1.776 (standard error 2.172) in the proportion of women in parliament following reserved seat quotas, and a reduction of 0.017 (standard error 0.071) in log(MMR).  Countries implementing candidate list quotas in the period under study are provided in notes to Figure \ref{fig:DIDMCandidates}.}
\end{figure}

\begin{figure}[htpb!]
  \centering
  \subfloat[Unweighted Estimates]{
    \includegraphics[scale=0.65]{./results/appendix/India/eventIndia.pdf}
  }
  \subfloat[Weighted Estimates]{
    \includegraphics[scale=0.65]{./results/appendix/India/eventIndiaWeighted.pdf}
  }
  
  \subfloat[Unweighted Estimates (with linear trends)]{
    \includegraphics[scale=0.65]{./results/appendix/India/eventIndiaTrends.pdf}
  }
  \subfloat[Weighted Estimates (with linear trends)]{
    \includegraphics[scale=0.65]{./results/appendix/India/eventIndiaWeightedTrends.pdf}
  }
  \caption{Event study analysis of reserved seats for women in large Indian states}
  \label{event-IndiaW}
  \floatfoot{\textsc{Notes:} Event studies are conducted using all available state-level estimates of maternal mortality in India, which are generated provided in the Office of the Registrar General \& Census Commissioner's Sample Registration System (SRS) Bulletins.  State level reforms refer to the reservation of seats for women in local councils, as described in \citet{Iyeretal2012}.  All standard errors are based on wild bootstrapped clustered standard errors (clustered by state), given the relatively low number of states.}
\end{figure}


\begin{landscape}
\begin{figure}[htpb!]
    \centering
    \includegraphics[scale=1.1]{./results/appendix/India/events_all.pdf}
    \caption{Leave-one-out event study estimates (India state reforms)}
    \label{LOOIndia}
    \floatfoot{\textsc{Notes:} Refer to notes to Figure \ref{event-IndiaW}.  Here identical event studies are estimated, however removing one state in each case on a leave-one-out basis. The left-out state is indicated in each subplot's title.}
\end{figure}
\end{landscape}

\begin{landscape}
\begin{figure}[htpb!]
    \caption{\citet{GoodmanBacon2018} decomposition based on 2$\times$2 difference-in-difference models}
    \label{fig:GB}
    \subfloat[Percent of women in parliament\label{GB-womparl}]{%
    \includegraphics[scale=1.2]{./results/appendix/GBdecomp_womparl_all.eps}%
    }
    \subfloat[ln(maternal mortality ratio)\label{GB-mmr}]{%
    \includegraphics[scale=1.2]{./results/appendix/GBdecomp_lnMMRt1_all.eps}%
    }
    \floatfoot{\textsc{Notes:} Figures document the Goodman-Bacon decomposition into a series of $2\times 2$ difference-in-differences models depending on the type of comparison unit. Plotted $\times$ symbols represent cases where identification is drawn from timing-only comparisons.  Darker shaded $\times$ symbols represent comparisons between earlier-treated units (as treatment) and later-treated units (as controls).  Lighter shaded $\times$ symbols represent (problematic) comparisons between later-treated units (as treatment) and earlier treated units (as controls).  Triangular symbols represent comparisons between treated (quota adopters) versus untreated pure controls (never adopters), with alternative estimates depending on the timing of adoption.  Finally, a small number of hollow circles represent comparisons between units which adopted quotas before the beginning of the panel versus units which later became treated.  Here each point on the graph considers an alternative adoption time period.  The global decomposition for each of these four groups is given in Table \ref{tab:GBglobal}.}
\end{figure}
\end{landscape}


\begin{figure}[htpp!]
  \subfloat[Afghanistan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_1.eps}%
  }
  \subfloat[Algeria]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_3.eps}%
  }
  \subfloat[Bangladesh]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_12.eps}%
  }
  \subfloat[Burundi]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_25.eps}%
  }

  \subfloat[China]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_33.eps}%
  }
  \subfloat[Djibouti]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_45.eps}%
  }
  \subfloat[Eritrea]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_51.eps}%
  }
  \subfloat[Haiti]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_68.eps}%
  }

  \subfloat[Iraq]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_75.eps}%
  }
  \subfloat[Jordan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_81.eps}%
  }
  \subfloat[Kenya]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_83.eps}%
  }
  \subfloat[Morocco]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_111.eps}%
  }

  \subfloat[Niger]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_119.eps}%
  }
  \subfloat[Pakistan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_123.eps}%
  }
  \subfloat[Rwanda]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_134.eps}%
  }
  \subfloat[Samoa]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_135.eps}%
  }

  \subfloat[Saudi Arabia]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_137.eps}%
  }
  \subfloat[South Sudan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_147.eps}%
  }
  \subfloat[Sudan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_152.eps}%
  }
  \subfloat[Swaziland]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_154.eps}%
  }

  \subfloat[Tanzania]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_159.eps}%
  }
  \subfloat[Uganda]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_168.eps}%
  }
  \subfloat[Zimbabwe]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_lnMMRt1_180.eps}%
  }
  \caption{Leave-one-out analysis: maternal mortality}
  \label{fig:DIDMLOO-MMR}
  \floatfoot{Notes: Plots replicate panel (b) of Figure \ref{fig:DIDM}, however leave out one quota-adopting country at a time.  The title of each plot refers to the country which is removed when estimating each $DID_M$ model.  All additional details follow the model presented in Figure \ref{fig:DIDM}.}
\end{figure}

\begin{figure}[htpp!]
  \subfloat[Afghanistan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_1.eps}%
  }
  \subfloat[Algeria]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_3.eps}%
  }
  \subfloat[Bangladesh]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_12.eps}%
  }
  \subfloat[Burundi]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_25.eps}%
  }

  \subfloat[China]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_33.eps}%
  }
  \subfloat[Djibouti]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_45.eps}%
  }
  \subfloat[Eritrea]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_51.eps}%
  }
  \subfloat[Haiti]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_68.eps}%
  }

  \subfloat[Iraq]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_75.eps}%
  }
  \subfloat[Jordan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_81.eps}%
  }
  \subfloat[Kenya]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_83.eps}%
  }
  \subfloat[Morocco]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_111.eps}%
  }

  \subfloat[Niger]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_119.eps}%
  }
  \subfloat[Pakistan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_123.eps}%
  }
  \subfloat[Rwanda]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_134.eps}%
  }
  \subfloat[Samoa]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_135.eps}%
  }

  \subfloat[Saudi Arabia]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_137.eps}%
  }
  \subfloat[South Sudan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_147.eps}%
  }
  \subfloat[Sudan]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_152.eps}%
  }
  \subfloat[Swaziland]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_154.eps}%
  }

  \subfloat[Tanzania]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_159.eps}%
  }
  \subfloat[Uganda]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_168.eps}%
  }
  \subfloat[Zimbabwe]{%
    \includegraphics[scale=0.38]{./results/appendix/loo/DIDM_womparl_180.eps}%
  }
  \caption{Leave-one-out analysis: women in parliament}
  \label{fig:DIDMLOO-WP}
  \floatfoot{Notes:   Plots replicate panel (a) of Figure \ref{fig:DIDM}, however leave out one quota-adopting country at a time.  The title of each plot refers to the country which is removed when estimating each $DID_M$ model.  All additional details follow the model presented in Figure \ref{fig:DIDM}.}
\end{figure}
    
    
\begin{figure}[htpp!]
  \subfloat[Afghanistan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_1.pdf}%
  }
  \subfloat[Algeria]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_3.pdf}%
  }
  \subfloat[Bangladesh]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_12.pdf}%
  }
  \subfloat[Burundi]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_25.pdf}%
  }

  \subfloat[China]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_33.pdf}%
  }
  \subfloat[Djibouti]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_45.pdf}%
  }
  \subfloat[Eritrea]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_51.pdf}%
  }
  \subfloat[Haiti]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_68.pdf}%
  }

  \subfloat[Iraq]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_75.pdf}%
  }
  \subfloat[Jordan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_81.pdf}%
  }
  \subfloat[Kenya]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_83.pdf}%
  }
  \subfloat[Morocco]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_111.pdf}%
  }

  \subfloat[Niger]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_119.pdf}%
  }
  \subfloat[Pakistan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_123.pdf}%
  }
  \subfloat[Rwanda]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_134.pdf}%
  }
  \subfloat[Samoa]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_135.pdf}%
  }

  \subfloat[Saudi Arabia]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_137.pdf}%
  }
  \subfloat[South Sudan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_147.pdf}%
  }
  \subfloat[Sudan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_152.pdf}%
  }
  \subfloat[Swaziland]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_154.pdf}%
  }

  \subfloat[Tanzania]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_159.pdf}%
  }
  \subfloat[Uganda]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_168.pdf}%
  }
  \subfloat[Zimbabwe]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_lnMMRt1_180.pdf}%
  }
  \caption{Leave-one-out analysis: maternal mortality}
  \label{fig:LOO-MMR}
  \floatfoot{Notes: Plots replicate panel (a) of Figure \ref{fig:eventReserved}, however leave out one quota-adopting country at a time.  The title of each plot refers to the country which is removed when estimating each DIDM study model.  All additional details follow the baseline model presented in Figure \ref{fig:eventReserved}.}
\end{figure}

\begin{figure}[htpp!]
  \subfloat[Afghanistan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_1.pdf}%
  }
  \subfloat[Algeria]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_3.pdf}%
  }
  \subfloat[Bangladesh]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_12.pdf}%
  }
  \subfloat[Burundi]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_25.pdf}%
  }

  \subfloat[China]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_33.pdf}%
  }
  \subfloat[Djibouti]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_45.pdf}%
  }
  \subfloat[Eritrea]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_51.pdf}%
  }
  \subfloat[Haiti]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_68.pdf}%
  }

  \subfloat[Iraq]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_75.pdf}%
  }
  \subfloat[Jordan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_81.pdf}%
  }
  \subfloat[Kenya]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_83.pdf}%
  }
  \subfloat[Morocco]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_111.pdf}%
  }

  \subfloat[Niger]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_119.pdf}%
  }
  \subfloat[Pakistan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_123.pdf}%
  }
  \subfloat[Rwanda]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_134.pdf}%
  }
  \subfloat[Samoa]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_135.pdf}%
  }

  \subfloat[Saudi Arabia]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_137.pdf}%
  }
  \subfloat[South Sudan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_147.pdf}%
  }
  \subfloat[Sudan]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_152.pdf}%
  }
  \subfloat[Swaziland]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_154.pdf}%
  }

  \subfloat[Tanzania]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_159.pdf}%
  }
  \subfloat[Uganda]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_168.pdf}%
  }
  \subfloat[Zimbabwe]{%
    \includegraphics[scale=0.25]{./results/appendix/loo/event_womparl_180.pdf}%
  }
  \caption{Leave-one-out analysis: women in parliament}
  \label{fig:LOO-WP}
  \floatfoot{Notes: Plots replicate panel (b) of Figure \ref{fig:eventReserved}, however leave out one quota-adopting country at a time.  The title of each plot refers to the country which is removed when estimating each event study model.  All additional details follow the baseline model presented in Figure \ref{fig:eventReserved}.}
\end{figure}

\begin{landscape}
\begin{figure}[htpp!]
\subfloat[Baseline model]{%
    \includegraphics[scale=0.5]{./results/appendix/ARpower_1.eps}%
}
\subfloat[GDP and democracy controls]{%
    \includegraphics[scale=0.5]{./results/appendix/ARpower_2.eps}%
}
\subfloat[Empowerment and predictor controls]{%
    \includegraphics[scale=0.5]{./results/appendix/ARpower_3.eps}%
}
\caption{Weak-IV Robust Rejection Levels}
\label{fig:weakIV}
\floatfoot{Notes: The graphs plot the rejection probabilities of null hypotheses estimated in IV models described in Table \ref{qIV}, where the horizontal axis plots the (hypothesised) value of the population parameter, and the vertical axis reports the rejection probability of the null that the population parameter is equal to the value on the horizontal axis, robust to weak instruments following Anderson-Rubin. The plot shows which parameters can be rejected at the 90\% level based on the A-R test-- any part of the red curve above the dashed black line is a null that can be rejected, even if the IV is weak. 
The Anderson Rubin confidence sets described in Table \ref{qIV} correspond to all values for which the null cannot be rejected (values for which the rejection curve falls below the dashed line). Plots are generated based on \citet{Finlayetal2013}.}
\end{figure}
\end{landscape}
  
\begin{landscape}
\begin{figure}[htpb!]
  \caption{\citeauthor{dCDH2020} estimates: women's rights and social standing from ``Varieties of Democracy'' data}
  \label{fig:DIDMWomenRights2}
  \subfloat[Women's Civil Liberties Index]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2x_gencl.eps}%
  }
  \subfloat[Women's Political Participation Index]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2x_genpp.eps}%
  }
  \subfloat[Exclusion by Gender Index]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2xpe_exlgender.eps}
  }
  
  \subfloat[Relative Freedom of Movement for Women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_moveRel.eps}
  }
  \subfloat[Relative Access to Justice for Women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_justRel.eps}%
  }
  \subfloat[Relative Freedom of Discussion for Women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_discRel.eps}%
  }

  \subfloat[Power distributed by gender]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2pepwrgen.eps}%
  }
  \subfloat[Freedom from forced labor for women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2clslavef.eps}%
  }
  \subfloat[Property rights for women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2clprptyw.eps}%
  }
  
    \vspace{-5mm}
    \floatfoot{\textsc{Notes:}
  \citet{dCDH2020} $DID_M$ estimates examine the variation of measures of women's rights and measures of relative social standing based on data collected in the Variety of Democracy dataset \citep{VDEM2020}. When generating ratios of women to male outcomes (in panels (d)-(f)) these are Winsorized at the 1\textsuperscript{st} and 99\textsuperscript{th} percentiles to avoid outliers in cases where the denominator is very small.  We follow suggested practices in removing a small number of observations which are coded by 3 or fewer country experts \citep{VDEM2020} in ``C'' type variables from the VDEM data which are based on the opinions of country experts.  Lead (placebo) and lag (dynamic) effects are estimated for each variable.} 
\end{figure}
\end{landscape}



\begin{landscape}    
\begin{figure}[htpb!]
  \caption{\citeauthor{dCDH2020} estimates: women's rights, empowerment, and women in politics}
  \label{fig:DIDMWomenRights}
  \subfloat[Women, Business \& Law Index (WB)]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_wblindex.eps}%
  }
  \subfloat[Women Ministers]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_womenMinisters.eps}%
  }
  \subfloat[Women's Protests]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_propWomenProtest.eps}%
  }

  
  \subfloat[CIRI Women's Social Rights]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_wosoc.eps}%
  }
  \subfloat[CIRI Women's Political Rights]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_wopol.eps}
  }
  \subfloat[CIRI Women's Economic Rights]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_wecon.eps}
  }
    
  \subfloat[Female Labor Force Participation]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_womenFLFP.eps}%
  }
  \subfloat[Abortion (Save Mother's Life)]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_tosavelifeofmother.eps}%
  }
  \subfloat[Abortion (Fetal Impairment)]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_fetalimpairment.eps}%
  }
    
  \vspace{-5mm}
  \floatfoot{\textsc{Notes:} 
  \citet{dCDH2020} $DID_M$ estimates are presented, with outcomes describing women's rights, empowerment or measures of participation in politics.  Panel (a) is a recently generated index provided by the World Bank capturing Women's participation in Business and Lawmaking. 
    Women Minister data is drawn from Inter-Parliamentary Union (IPU) Women in Politics data, women's protests as a share of all protests are calculated from data shared by \cite{bell2019impact}.
    Rights indexes are additionally defined by Cingranelli et al., and used in panels (d)-(f).  Female labor force participation is drawn from the World Development Indicators, and abortion laws are coded based on \citet{Eliasetal2017}.} 
\end{figure}
\end{landscape}



\begin{figure}[htpb!]
  \caption{\citeauthor{dCDH2020}'s $DID_{M}$ estimator and predictors from political science literature}
  \label{fig:dCDHPoliSci}
  \subfloat[Overseas Development Assistance p.c.]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_ODA.eps}%
  }
  \subfloat[Peace Keeper Presence]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_peacekeepers.eps}%
  }

  \vspace{-1mm}
  \subfloat[Right Wing Executive]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_execR.eps}
  }
  \subfloat[Left Wing Executive]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_execL.eps}
  }

  \vspace{-1mm}
  \subfloat[Years in Power]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_yrsoffc.eps}%
  }
  \subfloat[Herfindahl Index]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_herfgov.eps}%
  }

  \vspace{-1mm}
  \subfloat[Opposition's Vote Share]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_oppvote.eps}%
  }
  \subfloat[Regime in Transition]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_transition.eps}%
  }
  \vspace{-7mm}
  \floatfoot{\textsc{Notes:} \citet{dCDH2020} $DID_M$ estimates of potential quota predictors suggested in the political science literature are displayed.  All estimation details follow those described in notes to Figure \ref{fig:DIDM}.  Overseas Development Assistance (ODA) is measured as per capita net inflows in current US dollars, and is generated from the World Bank Data Bank.  Peacekeepers (measured in 1000s) are from the IPI Peacekeeping Database, and political measures including the orientation of leader's party, the leader's time in power, Herfindahl Index of parties, vote shares and regime types and changes are recorded by the Database of Political Institutions.  Indicators for the executive's political leaning are coded from the Database of Political Institutions, based on a classification of leaders into left (31.4\%), right (22.8\%), center (7.4\%) or not-applicable (38.4\%). 
  }
\end{figure}

\begin{figure}[htpb!]
\caption{DID$_M$ Estimates conditioning on potential quota predictors -- gender quota impacts on women in parliament and maternal mortality}
\label{fig:DIDMReservedPred}
\subfloat[Percent of women in parliament \label{DIDM-womparlPred}]{%
\includegraphics[scale=1]{./results/appendix/alternative/DIDM_womparl_AltInd.eps}%
}
\subfloat[ln(maternal mortality ratio) \label{DIDM-mmrPred}]{%
\includegraphics[scale=1]{./results/appendix/alternative/DIDM_lnMMRt1_AltInd.eps}%
}
\floatfoot{\textsc{Notes:} Plots present \citet{dCDH2020} $DID_M$ estimates replicating those in Figure \ref{fig:DIDM}, however now controlling for indexes constructed from  baseline measures of 7 potential predictors of quota timing from the political science literature (Figure \ref{fig:eventPoliSci}) and for 18 indicators of women's rights (Figures \ref{fig:eventWomenRights2}, \ref{fig:eventWomenRights}) interacted with post quota indicators. Two separate index$\times$post quota variables are constructed given different phenomena of interest: a first index considering quota predictors, and a second considering empowerment controls. Standard errors are based on a block bootstrap by country.} 
\end{figure}


\begin{figure}[htpb!]
\caption{Conditioning on potential quota predictors -- gender quota impacts on women in parliament and maternal mortality}
\label{fig:eventReservedPred}
\subfloat[Percent of women in parliament \label{event-womparlPred}]{%
\includegraphics[scale=0.52]{./results/appendix/alternative/empcontsPlusIndex_event_womparl_10.pdf}%
}
\subfloat[ln(maternal mortality ratio) \label{event-mmrPred}]{%
\includegraphics[scale=0.52]{./results/appendix/alternative/empcontsPlusIndex_event_lnMMRt1_10.pdf}%
}
\floatfoot{\textsc{Notes:} Event studies replicate those in Figure \ref{fig:eventReserved}, however now controlling for indexes based on baseline measures of 7 potential predictors of quota timing from the political science literature (Figure \ref{fig:eventPoliSci}) and for 18 indicators of women's rights (Figures \ref{fig:eventWomenRights2}, \ref{fig:eventWomenRights}) interacted with post quota indicators. Two separate index$\times$post quota variables are constructed given different phenomena of interest: a first index considering quota predictors, and a second considering empowerment controls.  Point estimates of the lag and lead terms in the event study specification described in equation \ref{eqn:wp-q} are presented, along with their 95\% confidence intervals. Estimates are conditional on country and year fixed effects. Time periods greater than 10 years from the reform date are displayed as a single ``10 +'' indicator. Standard errors are clustered by country. The omitted base category is taken as 1 year prior to the reform, indicated by the solid vertical line.} 
\end{figure}






\begin{figure}[htpp!]
  \subfloat[log(Fertility Rates)]{%
    \includegraphics[scale=1]{./results/appendix/DIDM_TFR1544.eps}%
  }
  \subfloat[Proportion of Girls]{%
    \includegraphics[scale=1]{./results/appendix/DIDM_sexMixDHS.eps}%
  }

  \subfloat[Mother's Education]{%
    \includegraphics[scale=1]{./results/appendix/DIDM_motherEduc.eps}%
  }
  \subfloat[Proportion of Illiterate Mothers]{%
    \includegraphics[scale=1]{./results/appendix/DIDM_motherIllit.eps}%
  }	
  \caption{Characteristics of births and mothers: $DID_M$ estimates from DHS birth pseudo-panel}
  \label{fig:DHScharacDIDM}
  \floatfoot{\textsc{Notes:} \citet{dCDH2020} $DID_M$ estimates are presented based on country by year averages generated from a pseudo-panel of births created from Demographic and Health Survey data.  Fertility rates refers to the log of the total fertility rate, calculated as the total number of births a woman would have if completing her fertile life based on current age-specific fertility rates. The proportion of girls refers to the proportion of girl births among all births.  Mother's education refers to the average years of education of mothers (women having given birth) in each country and year.  Proportion of illiterate mothers refers to the total proportion of all mothers (women having given birth) in each country and year who report not being able to read and write. All additional details follow notes to Figure \ref{fig:DIDM}.}
\end{figure}




\begin{figure}[htpp!]
  \subfloat[log(Fertility Rates)]{%
    \includegraphics[scale=0.5]{./results/appendix/event/event_TFR1544_10.pdf}%
  }
  \subfloat[Proportion of Girls]{%
    \includegraphics[scale=0.5]{./results/appendix/event/event_sexMixDHS_10.pdf}%
  }

  \subfloat[Mother's Education]{%
    \includegraphics[scale=0.5]{./results/appendix/event/event_motherEduc_10.pdf}%
  }
  \subfloat[Proportion of Illiterate Mothers]{%
    \includegraphics[scale=0.5]{./results/appendix/event/event_motherIllit_10.pdf}%
  }	
  \caption{Characteristics of births and mothers: Event study estimates from DHS birth pseudo-panel}
  \label{fig:DHScharac}
  \floatfoot{\textsc{Notes:} Event studies are estimated using country by year averages generated from a pseudo-panel of births created from Demographic and Health Survey data.  Fertility rates refers to the log of the total fertility rate, calculated as the total number of births a woman would have if completing her fertile life based on current age-specific fertility rates. The proportion of girls refers to the proportion of girl births among all births.  Mother's education refers to the average years of education of mothers (women having given birth) in each country and year.  Proportion of illiterate mothers refers to the total proportion of all mothers (women having given birth) in each country and year who report not being able to read and write.  All additional details follow notes to Figure \ref{fig:eventReserved}.}
\end{figure}
    
\begin{figure}[htpb!]
  \caption{\citeauthor{dCDH2020} estimates for impacts of gender quotas on age-specific birth rates calculated from DHS birth pseudo-panel}
  \label{fig:DHSDIDM}
  \subfloat[15--19]{%
    \includegraphics[scale=0.85]{./results/appendix/DIDM_birth1519.eps}%
  }
  \subfloat[20--24]{%
    \includegraphics[scale=0.85]{./results/appendix/DIDM_birth2024.eps}
  }

  \subfloat[25--29]{%
    \includegraphics[scale=0.85]{./results/appendix/DIDM_birth2529.eps}
  }
  \subfloat[30--34]{%
    \includegraphics[scale=0.85]{./results/appendix/DIDM_birth3034.eps}%
  }
  
  \subfloat[35--39]{%
    \includegraphics[scale=0.85]{./results/appendix/DIDM_birth3539.eps}%
  }
  \subfloat[40--44]{%
    \includegraphics[scale=0.85]{./results/appendix/DIDM_birth4044.eps}%
  }
  
  \floatfoot{Notes: \citet{dCDH2020}'s ``$DID_M$'' models are estimated using country by year averages generated from a pseudo-panel of births created from Demographic and Health Survey data.  Birth rates in each figure are calculated from birth indicators in the pseudo-panel based on women in each quinquennial age group in each country and year, where 1 indicates a surveyed woman (of this age) reported a birth in a given year, and 0 indicates they did not have a birth.} 
\end{figure}    


\begin{figure}[htpb!]
  \caption{Age-specific birth rates calculated from DHS birth pseudo-panel}
  \label{fig:DHS}
  \subfloat[15--19]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_birth1519_10.pdf}%
  }
  \subfloat[20--24]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_birth2024_10.pdf}
  }

  \subfloat[25--29]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_birth2529_10.pdf}
  }
  \subfloat[30--34]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_birth3034_10.pdf}%
  }
  
  \subfloat[35--39]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_birth3539_10.pdf}%
  }
  \subfloat[40--44]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_birth4044_10.pdf}%
  }
  
  \floatfoot{Notes: Event studies are estimated using country by year averages generated from a pseudo-panel of births created from Demographic and Health Survey data.  Birth rates in each figure are calculated from birth indicators in the pseudo-panel based on women in each quinquennial age group in each country and year, where 1 indicates a surveyed woman (of this age) reported a birth in a given year, and 0 indicates they did not have a birth.  Average birth rates across the full age range of 15--49 year-olds are documented in panel (a) of Figure \ref{fig:DHScharac}.} 
\end{figure}




\begin{figure}[htpb!]
\caption{DHS microdata -- gender quota impacts on women in parliament and MMR ($DID_M$ Estimates)}
\label{DIDM-MMRDHS}
\subfloat[Percent of women in parliament \label{didm-womparlDHS}]{%
\includegraphics[scale=1]{./results/appendix/DIDM_womparl_DHS.eps}%
}
\subfloat[ln(maternal mortality ratio)\label{didm-mmrDHS}]{%
 \includegraphics[scale=1]{./results/appendix/DIDM_lnMMR_DHS.eps}%
}
\floatfoot{\textsc{Notes:} Specification replicates Figure \ref{fig:DIDM}, however now replacing world-wide estimates of MMR with maternal mortality calculated from microdata reports from the DHS.}
\end{figure}

\begin{figure}[htpb!]
\caption{DHS microdata -- gender quota impacts on women in parliament and MMR (event study estimates)}
\label{event-MMRDHS}
\subfloat[Percent of women in parliament \label{event-womparl2way}]{%
\includegraphics[scale=0.52]{./results/appendix/event/Mechevent_womparl_DHS_level.pdf}%
}
\subfloat[ln(maternal mortality ratio) \label{event-mmr2way}]{%
\includegraphics[scale=0.52]{./results/appendix/event/Mechevent_lnMMR_DHS_level.pdf}%
}
\floatfoot{\textsc{Notes:} Specification replicates Figure \ref{fig:eventReserved}, however now replacing world-wide estimates of MMR with maternal mortality calculated from microdata reports from the DHS.
As the DHS maternal mortality module is available in only a subsample of the DHS countries (44 of 68 countries with publicly available surveys), we estimate using 2 year lags/leads to reduce noise. Substantively similar results obtain if using yearly lags and leads.}
\end{figure}



\begin{landscape}
\begin{figure}[htpb!]
  \caption{Mechanisms: event studies for impacts of gender quotas on intermediate outcomes based on a constant sample}
  \label{fig:eventMechanismBAL}
  \subfloat[Antenatal Care\label{event-antenatalBAL}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_antenatal_ipol_balanced.pdf}%
  }
  \subfloat[Attended Births\label{event-birthattendBAL}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_birthAttend_ipol_balanced.pdf}%
  }
  \subfloat[Modern Contraceptive Cover\label{eventmodernContBAL}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_contracepMod_balanced.pdf}
  }
  \vspace{-4mm}
  
  \subfloat[Fertility\label{eventBirthRateBAL}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_lnFert_balanced.pdf}%
  }
  \subfloat[Teen Pregnancy\label{eventBirthRateBAL}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_teenPreg_balanced.pdf}%
  }
  \subfloat[Women's Schooling (Average, 15-19)\label{eventGirlsEducBAL}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_yearsSchool1519_ipol_balanced.pdf}
  }
  \vspace{-4mm}
 
  \subfloat[Health Expenditure as a \% of GDP\label{event-healtExpBAL}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_healthExpt1_balanced.pdf}%
  }
  \subfloat[Development Assistance for Maternal Health\label{eventDAHBAL}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_p_mh_dah_17_balanced.pdf}
  }
  \subfloat[ln(GDP per capital)\label{eventGDPpcBAL}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_lnGDP_balanced.pdf}%
  }
  
  \floatfoot{\textsc{Notes:} Event study estimates of intermediate outcomes replicate those displayed in the paper in Figure \ref{fig:eventMechanism}.  However here all models work with a common sample containing observations for each intermediate outcome considered.  For additional notes, refer to Figure \ref{fig:eventMechanism}.} 
\end{figure}
\end{landscape}



\begin{figure}[htpp!]
  \subfloat[Democracy control only]{%
\includegraphics[scale=0.44]{./results/appendix/alternative/noGDP_event_lnMMRt1_10.pdf}%
    }
  \subfloat[GDP control only]{%
\includegraphics[scale=0.44]{./results/appendix/alternative/noDemoc_event_lnMMRt1_10.pdf}%
 }

  \subfloat[GDP and democracy controls]{%
    \includegraphics[scale=0.44]{./results/appendix/alternative/origcontrols_event_lnMMRt1_10.pdf}%
  }
  \subfloat[Demographic controls]{%
    \includegraphics[scale=0.44]{./results/appendix/alternative/moreconts_event_lnMMRt1_10.pdf}%
  }
  
  \subfloat[Region by Year FEs]{%
    \includegraphics[scale=0.44]{./results/appendix/alternative/regionFEs_event_lnMMRt1_10.pdf}%
    }
  \subfloat[Conditional on health expenditure]{%
    \includegraphics[scale=0.44]{./results/appendix/alternative/healthControls_event_lnMMRt1_10.pdf}%
  }
  
  \subfloat[Clustering by country and time -- ln(MMR)]{%
    \includegraphics[scale=0.44]{./results/appendix/alternative/twoway_event_lnMMRt1_10.pdf}%
  }
  \subfloat[Removing high income countries -- ln(MMR)]{%
    \includegraphics[scale=0.44]{./results/appendix/alternative/nohighinc_event_lnMMRt1_10.pdf}%
  }
  \caption{Alternative samples and specifications in quota event study (maternal mortality)}
  \label{fig:eventAlternativesQuota}
  \vspace{-8mm}
  \floatfoot{Notes: Each panel documents an alternative specification of the event study shown in Figure \ref{fig:eventReserved} panel (a). Specifications are shown with alternative controls, estimation samples, or modelling choics.  Demographic controls in panel (d) refer to time-varying controls of the education of fertile aged women, the proportion of the population in fertile ages, and the ethnic fractionalization of the population.  Region by year FEs include separate year fixed effects for every sub-region based on the United Nations classification.  Panel (f) additionally controls for health expenditure as a proportion of GDP, and Development Aid receipts for maternal health. Panel (g) estimates standard errors with double clustering (by country and year).  Panel (h)  removes high income countries from the control group.  A static (2015) measure of high income is used to ensure consistency of the sample across years.  Additional notes are provided in Figure \ref{fig:eventReserved}.}
\end{figure}

\begin{figure}[htpp!]
  \subfloat[Democracy control only]{%
\includegraphics[scale=0.46]{./results/appendix/alternative/noGDP_event_womparl_10.pdf}%
    }
  \subfloat[GDP control only]{%
\includegraphics[scale=0.46]{./results/appendix/alternative/noDemoc_event_womparl_10.pdf}%
 }

  \subfloat[GDP and democracy controls]{%
    \includegraphics[scale=0.46]{./results/appendix/alternative/origcontrols_event_womparl_10.pdf}%
  }
  \subfloat[Demographic controls]{%
    \includegraphics[scale=0.46]{./results/appendix/alternative/moreconts_event_womparl_10.pdf}%
  }
 
  \subfloat[Region by Year FEs]{%
    \includegraphics[scale=0.46]{./results/appendix/alternative/regionFEs_event_womparl_10.pdf}%
    }
  \subfloat[Conditional on health expenditure]{%
    \includegraphics[scale=0.46]{./results/appendix/alternative/healthControls_event_womparl_10.pdf}%
  }
  
 \subfloat[Clustering by country and time -- women in parliament]{%
    \includegraphics[scale=0.46]{./results/appendix/alternative/twoway_event_womparl_10.pdf}%
  }
  \subfloat[Removing high income countries -- women in parliament]{%
    \includegraphics[scale=0.46]{./results/appendix/alternative/nohighinc_event_womparl_10.pdf}%
  }
  \caption{Alternative samples and specifications in quota event study (women in parliament)}
  \label{fig:eventAlternativesQuotaWP}
  \floatfoot{Notes: Refer to notes to Figure \ref{fig:eventAlternativesQuota}.  Identical plots or shown where the outcome is women in parliament.}
\end{figure}


\begin{landscape}
\begin{figure}[htpb!]
  \caption{Mechanisms: Impacts of gender quotas on intermediate outcomes}
  \label{fig:eventMechanism}
  \subfloat[Antenatal care\label{event-antenatal}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_antenatal_ipol_10.pdf}%
  }
  \subfloat[Attended births\label{event-birthattend}]{%
\includegraphics[scale=0.45]{./results/appendix/event/event_birthAttend_ipol_10.pdf}%
  }
  \subfloat[Modern contraceptive cover\label{eventmodernCont}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_contracepMod_10.pdf}
  }
  \vspace{-4mm}
  
  \subfloat[ln(fertility)\label{eventBirthRate}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_lnFert_10.pdf}%
  }
  \subfloat[Teen pregnancy\label{eventTeenPreg}]{%
        \includegraphics[scale=0.45]{./results/appendix/event/event_teenPreg_10.pdf}%
  }
  \subfloat[Birth spacing\label{eventbirthSpace}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_birthSpacePost_10.pdf}
  }
  \vspace{-4mm}
  
  \subfloat[Health expenditure as a \% of GDP\label{event-healtExp}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_healthExpt1_10.pdf}%
  }
  \subfloat[Development assistance for maternal health\label{eventDAH}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_p_mh_dah_17_10.pdf}
  }
  \subfloat[ln(GDP per capital)\label{eventGDPpc}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_lnGDP_10.pdf}%
  }
  \vspace{-6mm}
  \floatfoot{\textsc{Notes:} Event-study estimates of intermediate outcomes replicate those from \ref{fig:DIDMMechanism}, however now following specification \ref{eqn:wp-q}. Estimates in Panels A and B use a linearly interpolated measure, event studies based on non-imputed values are available in Appendix Figure \ref{fig:mech2y}.  All other details follow those of event studies in Figure \ref{fig:eventReserved}, and variable definitions in Figure \ref{fig:DIDMMechanism}.} 
\end{figure}
\end{landscape}


\begin{landscape}
\begin{figure}[htpb!]
  \caption{Event study estimates: Women's rights and social standing from ``Varieties of Democracy'' data}
  \label{fig:eventWomenRights2}
  \subfloat[Women's Civil Liberties Index]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_v2x_gencl_10.pdf}%
  }
  \subfloat[Women's Political Participation Index]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_v2x_genpp_10.pdf}%
  }
  \subfloat[Exclusion by Gender Index]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_v2xpe_exlgender_10.pdf}
  }

  \subfloat[Relative Freedom of Movement for Women]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_moveRel_10.pdf}
  }
  \subfloat[Relative Access to Justice for Women]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_justRel_10.pdf}%
  }
  \subfloat[Relative Freedom of Discussion for Women]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_discRel_10.pdf}%
  }

  \subfloat[Power distributed by gender]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_v2pepwrgen_10.pdf}%
  }
  \subfloat[Freedom from forced labor for women]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_v2clslavef_10.pdf}%
  }
  \subfloat[Property rights for women]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_v2clprptyw_10.pdf}%
  }
  
  \floatfoot{\textsc{Notes:} Refer to Notes to Figure \ref{fig:DIDMWomenRights2}.  Here models based on identical outcome variables are estimated, however using the standard event study implementation described in equation \ref{eqn:wp-q}.} 
\end{figure}
    \end{landscape}



 
\begin{landscape}
\begin{figure}[htpb!]
  \caption{Event study estimates: Women's rights, empowerment, and women in politics}
  \label{fig:eventWomenRights}
  \subfloat[Women, Business \& Law Index (WB)]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_wblindex_10.pdf}%
  }
  \subfloat[Women Ministers]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_womenMinisters_10.pdf}%
  }
  \subfloat[Women's Protests]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_propWomenProtest_10.pdf}%
  }
  
  \subfloat[CIRI Women's Social Rights]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_wosoc_10.pdf}%
  }
  \subfloat[CIRI Women's Political Rights]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_wopol_10.pdf}
  }
  \subfloat[CIRI Women's Economic Rights]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_wecon_10.pdf}
  }

  \subfloat[Female Labor Force Participation]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_womenFLFP_10.pdf}%
  }
  \subfloat[Abortion (Save Mother's Life)]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_tosavelifeofmother_10.pdf}%
  }
  \subfloat[Abortion (Fetal Impairment)]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_fetalimpairment_10.pdf}%
  }

  \vspace{-2mm}
  \floatfoot{\textsc{Notes:} Refer to Notes to Figure \ref{fig:DIDMWomenRights}. Here models based on identical outcome variables are estimated, however using the standard event study estimator following equation \ref{eqn:wp-q}.} 
\end{figure}
\end{landscape}








        






\begin{figure}[htpb!]
  \caption{\citeauthor{RambachanRoth2019} estimates: women's rights and social standing from ``Varieties of Democracy'' data}
  \label{fig:RRWomenRights2}
  \subfloat[Women's Civil Liberties Index]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_v2x_gencl.eps}%
  }
  \subfloat[Women's Political Participation Index]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_v2x_genpp.eps}%
  }
  \subfloat[Exclusion by Gender Index]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_v2xpe_exlgender.eps}
  }
  
  \subfloat[Relative Freedom of Movement for Women]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_moveRel.eps}
  }
  \subfloat[Relative Access to Justice for Women]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_justRel.eps}%
  }
  \subfloat[Relative Freedom of Discussion for Women]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_discRel.eps}%
  }

  \subfloat[Power distributed by gender]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_v2pepwrgen.eps}%
  }
  \subfloat[Freedom from forced labor for women]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_v2clslavef.eps}%
  }
  \subfloat[Property rights for women]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_v2clprptyw.eps}%
  }
  
  \floatfoot{\textsc{Notes:} Refer to Notes to Figure \ref{fig:eventWomenRights2}.  Here models based on identical outcome variables are estimated, however in this case, each post-quota coefficient from event study \ref{eqn:wp-q} is documented, along with valid inference under \citet{RambachanRoth2019}'s ``Honest DiD'' methods.  Additional notes are available in Figure \ref{fig:HonestMain}.} 
\end{figure}


\begin{figure}[htpb!]
  \caption{\citeauthor{RambachanRoth2019} estimates: women's rights, empowerment, and women in politics}
  \label{fig:RRWomenRights}
  \subfloat[Women, Business \& Law Index (WB)]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_wblindex.eps}%
  }
  \subfloat[Women Ministers]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_womenMinisters.eps}%
  }
  \subfloat[Women's Protests]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_propWomenProtest.eps}%
  }

  
  \subfloat[CIRI Women's Social Rights]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_wosoc.eps}%
  }
  \subfloat[CIRI Women's Political Rights]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_wopol.eps}
  }
  \subfloat[CIRI Women's Economic Rights]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_wecon.eps}
  }
    
  \subfloat[Female Labor Force Participation]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_womenFLFP.eps}%
  }
  \subfloat[Abortion (Save Mother's Life)]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_tosavelifeofmother.eps}%
  }
  \subfloat[Abortion (Fetal Impairment)]{%
    \includegraphics[scale=0.35]{./results/appendix/HonestDiD/EventHonest_fetalimpairment.eps}%
  }
    
  \vspace{-2mm}
  \floatfoot{\textsc{Notes:} Refer to Figure \ref{fig:eventWomenRights}. Here models based on identical outcome variables are estimated, however in this case, each post-quota coefficient from event study \ref{eqn:wp-q} is documented, along with valid inference under \citet{RambachanRoth2019}'s ``Honest DiD'' methods.  Additional notes are available in Figure \ref{fig:HonestMain}.} 
\end{figure}

\begin{landscape}
\begin{figure}[htpb!]
  \caption{Mechanisms: \citeauthor{dCDH2020} estimates for impacts of gender quotas on intermediate outcomes with country-specific linear trends}
  \label{fig:DIDMMechanismtrend}
  \subfloat[Antenatal care\label{DIDM-antenatalT}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_antenatal_ipol_trend.eps}%
  }
  \subfloat[Attended births\label{DIDM-birthattendT}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_birthAttend_ipol_trend.eps}%
  }
  \subfloat[Modern contraceptive cover\label{DIDM-modernContT}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_contracepMod_trend.eps}
  }
  \vspace{-4mm}
  
  \subfloat[ln(fertility)\label{DIDM-BirthRateT}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_lnFert_trend.eps}%
  }
  \subfloat[Teen pregnancy\label{DIDM-BirthRateTeenT}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_teenPreg_trend.eps}%
  }
  \subfloat[Birth spacing\label{DIDM-birthSpaceT}]{%
\includegraphics[scale=0.72]{./results/appendix/DIDM_birthSpacePost_trend.eps}
  }
  \vspace{-4mm}
  
  \subfloat[Health expenditure as a \% of GDP\label{DIDM-healtExpT}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_healthExpt1_trend.eps}%
  }
  \subfloat[Development assistance for maternal health\label{DIDM-DAHT}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_p_mh_dah_17_trend.eps}
  }
  \subfloat[ln(GDP per capital)\label{DIDM-GDPpcT}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_lnGDP_trend.eps}%
  }
  \vspace{-1cm}
  \floatfoot{\textsc{Notes:} Refer to notes to Figure \ref{fig:DIDMMechanism}. Identical models are estimated, however here including country-specific linear trends over time.} 
\end{figure}
\end{landscape}

\begin{landscape}
\begin{figure}[htpb!]
  \caption{\citeauthor{dCDH2020} estimates with country-specific time trends: women's rights and social standing from ``Varieties of Democracy'' data}
  \label{fig:DIDMWomenRights2trend}
  \subfloat[Women's Civil Liberties Index]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2x_gencl_trend.eps}%
  }
  \subfloat[Women's Political Participation Index]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2x_genpp_trend.eps}%
  }
  \subfloat[Exclusion by Gender Index]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2xpe_exlgender_trend.eps}
  }
  
  \subfloat[Relative Freedom of Movement for Women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_moveRel_trend.eps}
  }
  \subfloat[Relative Access to Justice for Women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_justRel_trend.eps}%
  }
  \subfloat[Relative Freedom of Discussion for Women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_discRel_trend.eps}%
  }

  \subfloat[Power distributed by gender]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2pepwrgen_trend.eps}%
  }
  \subfloat[Freedom from forced labor for women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2clslavef_trend.eps}%
  }
  \subfloat[Property rights for women]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_v2clprptyw_trend.eps}%
  }
  
    \vspace{-5mm}
    \floatfoot{\textsc{Notes:} 
  Refer to notes to Figure \ref{fig:DIDMWomenRights2}. Identical models are estimated, however here including country-specific linear trends over time.}
\end{figure}
\end{landscape}

\begin{landscape}    
\begin{figure}[htpb!]
  \caption{\citeauthor{dCDH2020} estimates with country-specific time trends: women's rights, empowerment, and women in politics}
  \label{fig:DIDMWomenRightstrends}
  \subfloat[Women, Business \& Law Index (WB)]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_wblindex_trend.eps}%
  }
  \subfloat[Women Ministers]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_womenMinisters.eps}% DC CHECK 2022
  }
  \subfloat[Women's Protests]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_propWomenProtest_trend.eps}%
  }

  
  \subfloat[CIRI Women's Social Rights]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_wosoc_trend.eps}%
  }
  \subfloat[CIRI Women's Political Rights]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_wopol_trend.eps}
  }
  \subfloat[CIRI Women's Economic Rights]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_wecon_trend.eps}
  }
    
  \subfloat[Female Labor Force Participation]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_womenFLFP_trend.eps}%
  }
  \subfloat[Abortion (Save Mother's Life)]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_tosavelifeofmother_trend.eps}%
  }
  \subfloat[Abortion (Fetal Impairment)]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_fetalimpairment_trend.eps}%
  }
    
  \vspace{-5mm}
  \floatfoot{\textsc{Notes:} Refer to notes to Figure \ref{fig:DIDMWomenRights}. Identical models are estimated, however here including country-specific linear trends over time.} 
\end{figure}
\end{landscape}  


\begin{landscape}
  \begin{figure}[htpb!]
    \subfloat[Quotas and Democracy]{
    \includegraphics[scale=0.45]{./results/main/event_democracy_10.pdf}
    }%
    \subfloat[Democracy and Maternal Mortality]{%
    \includegraphics[scale=0.45]{./results/main/event_lnMMRt1_democracy_10.pdf}
    }%  
    \subfloat[Quotas and MMR, conditional on Democracy]{
    \includegraphics[scale=0.45]{./results/main/event_lnMMRt1_democracyCOND_10.pdf}
    }
    \caption{Quotas, Democratic Transitions and Maternal Mortality Declines -- Event study estimates}
    \label{fig:democracyTests2}
    \floatfoot{\textsc{Notes:} Here we consider the potential role of democratization in observed impacts of quota adoption on maternal mortality. Left-hand panel estimates the impact of quota adoption on whether or not a country is classified as a democracy.  Center panel estimates the impact of transition to a democracy (rather than quota adoption) on rates of maternal mortality.  Right-hand panel reports the impacts of quota adoption on maternal mortality when additionally including as controls full lags and leads to the adoption of democracy in cases where countries are classified as moving from non-democratic to democratic.  All other estimation details follow those in Figure \ref{fig:eventReserved}.}
    \end{figure}
\end{landscape}


    
\clearpage

 

\begin{landscape}
\begin{figure}[htpp!]
  \subfloat[Tuberculosis mortality\label{DIDMTBMort}]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_invsinTBt1.eps}%
  }
  \subfloat[Male adult mortality\label{DIDMMaleMort}]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_lnmmortt1.eps}%
  }
  \subfloat[Female adult mortality\label{DIDMFemaleMort}]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_lnfmortt1.eps}%
  }

  \subfloat[Infant Mortality\label{DIDMTBMort}]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_lnIMR_DHS.eps}%
  }
  \subfloat[Male Infant Mortality\label{DIDMMaleInfMort}]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_lnIMR_m_DHS.eps}%
  }
  \subfloat[Female Infant Mortality\label{DIDMFemaleInfMort}]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_lnIMR_f_DHS.eps}%
  }
  \caption{\citet{dCDH2020} $DID_M$ estimates for the impact of gender quotas on alternative health outcomes}
  \label{fig:DIDMGenderNeut}
  \floatfoot{\textsc{Notes:} Identical DID$_M$ estimates are plotted to those in Figure \ref{fig:DIDM}, however examining quota impacts on alternative health outcomes.  These outcomes are the rate of death due to TB per 100,000 individuals, and infant mortality per 1,000 live births, male mortality per 1,000 adult males (ages 15--60), and female mortality per 1,000 adult females (ages 15--60). For comparison with Figure \ref{event-mmr}, the natural logarithm of each variable is used, with the exception of TB mortality, where an inverse sine transformation is used, given a small number of country-year observations where a rate of zero is observed.  Infant mortality is recorded from Demographic and Health Survey (DHS) microdata where mothers report their full fertility history, any children who have died, their child's age at death (if relevant) and the year of death (if relevant).  We generate infant mortality rates using retrospective fertility and survival histories for DHS surveys in the 68 publicly available DHS countries.}
  \end{figure}
\end{landscape}


\begin{landscape}
\begin{figure}[htpp!]
  \subfloat[Tuberculosis mortality\label{eventTBMort}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_invsinTBt1_10.pdf}%
  }
  \subfloat[Male adult mortality\label{eventMaleMort}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_lnmmortt1_10.pdf}%
  }
  \subfloat[Female adult mortality\label{eventFemaleMort}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_lnfmortt1_10.pdf}%
  }

  \subfloat[Infant Mortality\label{eventTBMort}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_lnIMR_DHS_10.pdf}%
  }
  \subfloat[Male Infant Mortality\label{eventMaleInfMort}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_lnIMR_m_DHS_10.pdf}%
  }
  \subfloat[Female Infant Mortality\label{eventFemaleInfMort}]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_lnIMR_f_DHS_10.pdf}%
  }
  \caption{Event Studies for the impact of gender quotas on alternative health outcomes}
  \label{fig:eventGenderNeut}
  \floatfoot{\textsc{Notes:} Identical event studies are plotted to those in Figure \ref{fig:eventReserved}, however examining quota impacts on alternative health outcomes.  These outcomes are the rate of death due to TB per 100,000 individuals, and infant mortality per 1,000 live births, male mortality per 1,000 adult males (ages 15--60), and female mortality per 1,000 adult females (ages 15--60). For comparison with Figure \ref{event-mmr}, the natural logarithm of each variable is used, with the exception of TB mortality, where an inverse sine transformation is used, given a small number of country-year observations where a rate of zero is observed.  Infant mortality is recorded from Demographic and Health Survey (DHS) microdata where mothers report their full fertility history, any children who have died, their child's age at death (if relevant) and the year of death (if relevant).  We generate infant mortality rates using retrospective fertility and survival histories for DHS surveys in the 68 publicly available DHS countries.}
  \end{figure}
\end{landscape}




    
\begin{figure}[htpb!]
  \caption{Health spending -- various $DID_M$ estimates}
  \label{fig:healthSpendingDIDM}
  \subfloat[Original model with no controls]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_healthExpt1.eps}%
  }\hspace{2mm}
  \subfloat[Original model adding controls]{%
    \includegraphics[scale=0.7]{./results/appendix/alternative/origcontrols_DIDM_healthExpt1.eps}%
  }
  
  \subfloat[Extrapolating to cover 1990--1994]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_healthExpt1_ipol.eps}
  }\hspace{2mm}
  \subfloat[Imputing 1990--1994 with 1995]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_healthExpReplace.eps}
  }

  \subfloat[Proportion of Aid to Maternal Health]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_maternalAidShare.eps}%
  }\hspace{2mm}
  \subfloat[Maternal Health Aid p.c.]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_maternalAidPC.eps}%
  }

  \subfloat[ln(Government Health Spending p.c.)]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_logHealthSpend.eps}%
  }\hspace{2mm}
  \subfloat[ln(Gov't Health Spend) Imputing `90--`99 with 2000]{%
    \includegraphics[scale=0.7]{./results/appendix/DIDM_logHealthReplace.eps}%
  }
  
  \vspace{-8mm}
  \floatfoot{\textsc{Notes:} Alternative $DID_M$ estimates are displayed examining the impact of quota reform on health spending.  A baseline model is displayed in Figure \ref{fig:DIDMMechanism} (replicated in panel (a) here).  Health spending as a proportion of GDP is available world-wide from 1995 onwards.  Panels (a) and (b) use original data removing 1990--1994 from the sample.  Panels (c) and (d) linearly extrapolate to cover 1990--1994, or use values from 1995 to impute for 1990--1994.  Panels (e)--(g) use alternative measures of health spending (proportion of all overseas aid receipts for health dedicate to maternal health, maternal health aid per capita, or the log of total government health spending per capita.  The log of government health spending per capita in panel (g) is only measured from 2000 onwards, and so is subset to the period of 2000 and later.  Panel (h) re-estimates panel (g), however additionally imputing values from 1990-1999 with the value for each country in 2000.} 
\end{figure}
\clearpage    
    
\begin{figure}[htpb!]
  \caption{Health spending -- various event study specifications}
  \label{fig:healthSpending}
  \subfloat[Original model with no controls]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_healthExpt1_10.pdf}%
  }
  \subfloat[Original model adding controls]{%
    \includegraphics[scale=0.45]{./results/auxiliary/origcontrols_event_healthExpt1_10.pdf}%
  }
  
  \subfloat[Extrapolating to cover 1990--1994]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_healthExpt1_ipol_10.pdf}
  }
  \subfloat[Imputing 1990--1994 with 1995]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_healthExpReplace_10.pdf}
  }

  \subfloat[Proportion of Aid to Maternal Health]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_maternalAidShare_10.pdf}%
  }
  \subfloat[Maternal Health Aid p.c.]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_maternalAidPC_10.pdf}%
  }

  \subfloat[ln(Government Health Spending p.c.)]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_logHealthSpend_10.pdf}%
  }
  \subfloat[ln(Gov't Health Spend) Imputing 1990--`99 with 2000]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_logHealthReplace_10.pdf}%
  }
  
  \vspace{-2mm}
  \floatfoot{\textsc{Notes:} Alternative event-study models are displayed examining the impact of quota reform on health spending.  A baseline model is displayed in Figure \ref{fig:eventMechanism} (replicated in panel (a) here), and Honest DiD lags are presented in Figure \ref{fig:honestMechanism}.  Health spending as a proportion of GDP is available world-wide from 1995 onwards.  Panels (a) and (b) use original data removing 1990--1994 from the sample.  Panels (c) and (d) linearly extrapolate to cover 1990--1994, or use values from 1995 to impute for 1990--1994.  Panels (e)--(g) use alternative measures of health spending (proportion of all overseas aid receipts for health dedicate to maternal health, maternal health aid per capita, or the log of total government health spending per capita. The log of government health spending per capita in panel (g) is only measured from 2000 onwards, and so is subset to the period of 2000 and later.  Panel (h) re-estimates panel (g), however additionally imputing values from 1990-1999 with the value for each country in 2000.
  } 
\end{figure}

\begin{figure}[htpb!]
  \caption{Event studies: reserved seat quotas and predictors from political science literature}
  \label{fig:eventPoliSci}
  \subfloat[Overseas Development Assistance p.c.]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_ODA_10.pdf}%
  }
  \subfloat[Peace Keeper Presence]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_peacekeepers_10.pdf}%
  }

  \subfloat[Right Wing Executive]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_execR_10.pdf}
  }
  \subfloat[Left Wing Executive]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_execL_10.pdf}
  }

  \subfloat[Years in Power]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_yrsoffc_10.pdf}%
  }
  \subfloat[Herfindahl Index]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_herfgov_10.pdf}%
  }

  \subfloat[Opposition's Vote Share]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_oppvote_10.pdf}%
  }
  \subfloat[Regime in Transition]{%
    \includegraphics[scale=0.45]{./results/appendix/event/event_transition_10.pdf}%
  }

  \floatfoot{\textsc{Notes:} Figure replicates models of variation in quota predictors as flagged in the political science literature presented in Figure \ref{fig:dCDHPoliSci}, however now estimating  based on standard event studies.  Refer to notes to Figure \ref{fig:dCDHPoliSci} for variable definitions.} 
\end{figure}




\clearpage

 

\begin{landscape}
\begin{figure}
    \centering
    \caption{Synthetic Difference-in-Differences, Outcome Trends and Synthetic Weights}
    \label{fig:SDIDlnMMR}
    \subfloat[Quota Year 2000 (trends)]{%
      \includegraphics[scale=0.4]{./results/appendix/SDID_lnMMRtrends2000.eps}%
    }
    \subfloat[Quota Year 2002 (trends)]{%
      \includegraphics[scale=0.4]{./results/appendix/SDID_lnMMRtrends2002.eps}%
    }
    \subfloat[Quota Year 2003 (trends)]{%
      \includegraphics[scale=0.4]{./results/appendix/SDID_lnMMRtrends2003.eps}%
    }
    \subfloat[Quota Year 2005 (trends)]{%
      \includegraphics[scale=0.4]{./results/appendix/SDID_lnMMRtrends2005.eps}%
    }

    \subfloat[Quota Year 2000 (weights)]{%
      \includegraphics[scale=0.7]{./results/appendix/SDID_lnMMRweights2000.eps}%
    }
    \subfloat[Quota Year 2002 (weights)]{%
      \includegraphics[scale=0.7]{./results/appendix/SDID_lnMMRweights2002.eps}%
    }
    \subfloat[Quota Year 2003 (weights)]{%
      \includegraphics[scale=0.7]{./results/appendix/SDID_lnMMRweights2003.eps}%
    }
    \subfloat[Quota Year 2005 (weights)]{%
      \includegraphics[scale=0.7]{./results/appendix/SDID_lnMMRweights2005.eps}%
    }
    \floatfoot{\textsc{Notes:} Adoption-specific synthetic difference-in-differences estimates are displayed following \citet{arkhangelsky2021synthetic}.  Top panels display average outcomes for maternal mortality in treated and synthetic control units, along with time-specific weights calculated according to \citet{arkhangelsky2021synthetic}. Lower panels display weights assigned to each untreated country (size of points), as well as state-by-state observed differences calculated as $\hat\delta_{tr}-\hat\delta_{i}$ following \citet{arkhangelsky2021synthetic}, and the vertical dotted line presents the weighted averages of these differences (the estimated effect).  Observations with 0 weight are denoted using an x symbol. Full synthetic DID estimates are based on weighted averages of adoption year specific estimates.  In the interests of space, here we only present the first four adoption years (adoption years 2010, 2012, and 2013 are not displayed). Implementation follows \citep{PailanirClarke2022}.}
\end{figure}   
\end{landscape}







    

    

\begin{figure}[htpp!]
  \subfloat[Democracy control only]{%
\includegraphics[scale=0.68]{./results/appendix/alternative/noGDP_DIDM_lnMMRt1.eps}%
    }
  \subfloat[GDP control only]{%
\includegraphics[scale=0.68]{./results/appendix/alternative/noDemoc_DIDM_lnMMRt1.eps}%
 }

  \subfloat[GDP and democracy controls]{%
    \includegraphics[scale=0.68]{./results/appendix/alternative/origcontrols_DIDM_lnMMRt1.eps}%
  }
  \subfloat[Demographic controls]{%
    \includegraphics[scale=0.68]{./results/appendix/alternative/moreconts_DIDM_lnMMRt1.eps}%
  }
  
  \subfloat[Region by Year FEs]{%
    \includegraphics[scale=0.68]{./results/appendix/alternative/regionFEs_DIDM_lnMMRt1.eps}%
    }
  \subfloat[Conditional on health expenditure]{%
    \includegraphics[scale=0.68]{./results/appendix/alternative/healthControls_DIDM_lnMMRt1.eps}%
  }
  
  \subfloat[Removing high income countries -- ln(MMR)]{%
    \includegraphics[scale=0.78]{./results/appendix/alternative/nohighinc_DIDM_lnMMRt1.eps}%
  }
  \caption{Alternative samples and specifications in DID$_M$ models (maternal mortality)}
  \label{fig:DIDMAlternativesQuota}
  \vspace{-8mm}
  \floatfoot{Notes: Each panel documents an alternative specification of \citet{dCDH2020}'s DID$_M$ estimates, as displayed in  Figure \ref{fig:DIDM}. Specifications are shown with alternative controls, estimation samples, or modelling choics.  Demographic controls in panel (d) refer to time-varying controls of the education of fertile aged women, the proportion of the population in fertile ages, and the ethnic fractionalization of the population.  Region by year FEs include separate year fixed effects for every sub-region based on the United Nations classification.  Panel (f) additionally controls for health expenditure as a proportion of GDP, and Development Aid receipts for maternal health. Panel (h)  removes high income countries from the control group.  A static (2015) measure of high income is used to ensure consistency of the sample across years.  Additional notes are provided in Figure \ref{fig:DIDM}.}
\end{figure}

\begin{figure}[htpp!]
  \subfloat[Democracy control only]{%
\includegraphics[scale=0.68]{./results/appendix/alternative/noGDP_DIDM_womparl.eps}%
    }
  \subfloat[GDP control only]{%
\includegraphics[scale=0.68]{./results/appendix/alternative/noDemoc_DIDM_womparl.eps}%
 }

  \subfloat[GDP and democracy controls]{%
    \includegraphics[scale=0.68]{./results/appendix/alternative/origcontrols_DIDM_womparl.eps}%
  }
  \subfloat[Demographic controls]{%
    \includegraphics[scale=0.68]{./results/appendix/alternative/moreconts_DIDM_womparl.eps}%
  }
 
  \subfloat[Region by Year FEs]{%
    \includegraphics[scale=0.68]{./results/appendix/alternative/regionFEs_DIDM_womparl.eps}%
    }
  \subfloat[Conditional on health expenditure]{%
    \includegraphics[scale=0.68]{./results/appendix/alternative/healthControls_DIDM_womparl.eps}%
  }
  
  \subfloat[Removing high income countries -- women in parliament]{%
    \includegraphics[scale=0.78]{./results/appendix/alternative/nohighinc_DIDM_womparl.eps}%
  }
  \caption{Alternative samples and specifications in DID$_M$ models (women in parliament)}
  \label{fig:DIDMAlternativesQuotaWP}
  \floatfoot{Notes: Refer to notes to Figure \ref{fig:DIDMAlternativesQuota}.  Identical plots or shown where the outcome is women in parliament.}
\end{figure}







\begin{landscape}
\begin{figure}[htpb!]
  \caption{Mechanisms: $DID_M$ estimates for impacts of gender quotas on intermediate outcomes based on a constant sample}
  \label{fig:DIDMMechanismBAL}
  \subfloat[Antenatal Care\label{DIDM-antenatalBAL}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_antenatal_ipol_balanced.eps}%
  }
  \subfloat[Attended Births\label{DIDM-birthattendBAL}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_birthAttend_ipol_balanced.eps}%
  }
  \subfloat[Modern Contraceptive Cover\label{DIDMmodernContBAL}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_contracepMod_balanced.eps}
  }
  \vspace{-4mm}
  
  \subfloat[Fertility\label{DIDMBirthRateBAL}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_lnFert_balanced.eps}%
  }
  \subfloat[Teen Pregnancy\label{DIDMBirthRateBAL}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_teenPreg_balanced.eps}%
  }
  \subfloat[Women's Schooling (Average, 15-19)\label{DIDMGirlsEducBAL}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_yearsSchool1519_ipol_balanced.eps}
  }
  \vspace{-4mm}
 
  \subfloat[Health Expenditure as a \% of GDP\label{DIDM-healtExpBAL}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_healthExpt1_balanced.eps}%
  }
  \subfloat[Dev.\ Assistance for Maternal Health\label{DIDMDAHBAL}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_p_mh_dah_17_balanced.eps}
  }
  \subfloat[ln(GDP per capital)\label{DIDMGDPpcBAL}]{%
    \includegraphics[scale=0.72]{./results/appendix/DIDM_lnGDP_balanced.eps}%
  }
  
  \floatfoot{\textsc{Notes:} \citet{dCDH2020} $DID_M$ estimates of intermediate outcomes replicate those displayed in the paper in Figure \ref{fig:DIDMMechanism}.  However here all models work with a common sample containing observations for each intermediate outcome considered.  For additional notes, refer to Figure \ref{fig:DIDMMechanism}.} 
\end{figure}
\end{landscape}



\clearpage

    
\begin{landscape}
\begin{figure}[htpb!]
    \caption{Mechanisms: event studies for impacts of quotas on non-interpolated antenatal care and birth attendance}
    \label{fig:mech2y}
    \subfloat[Antenatal Care Coverage\label{event-antenatal2y}]{%
    \includegraphics[scale=0.68]{./results/appendix/event/Mechevent_antenatalt1_level.pdf}%
    }
    \subfloat[Attended Births\label{event-birthatt2y}]{%
    \includegraphics[scale=0.68]{./results/appendix/event/Mechevent_birthAttendt1_level.pdf}%
    }
    \floatfoot{\textsc{Notes:} Event studies replicate panels (a) and (b) of Figure \ref{fig:eventMechanism} using non-interpolated data. Given the unbalanced coverage of antenatal care and attended birth measures by countries and years, we present estimates for these outcomes pooling in 2 yearly bins, rather than yearly bins, to avoid unbalanced coverage in particular lag and lead terms where possible.}
\end{figure}
\end{landscape}


\clearpage
\section{Appendix Tables}    




\begin{table}[htpb!]
  \caption{Summary statistics for reserved seat analysis}
  \label{ssquota}
  \begin{tabular}{lccccc} \toprule
    & Obs.\ & Mean & Std.\ Dev.\  & Min.\ & Max.\ \\ \midrule
    \multicolumn{6}{l}{\textbf{Panel A: Full Sample}} \\
    \input{./results/appendix/summary/SumQuota.tex}
    \multicolumn{6}{l}{\textbf{Panel B: Reserved Seat Sample}} \\
    \input{./results/appendix/summary/SumQuota_Res.tex}
    \multicolumn{6}{l}{\textbf{Panel C: No Reserved Seat Sample}} \\
    \input{./results/appendix/summary/SumQuota_NoRes.tex}
    \bottomrule
    \multicolumn{6}{p{15cm}}{{\footnotesize \textsc{Notes:} Refer to Data Appendix \ref{app:Data} for a full description of each variable and its source.  The Maternal Mortality Ratio is measured as deaths per 100,000 live births.  For comparison, the male and female mortality rates for 15--60 year-olds is expressed as per 1,000 male and female adults respectively. Reserved seats is a binary variable taking one for each country and year pair where a quota was implemented, and 0 otherwise.}}
\end{tabular}
\end{table}


\clearpage
\definecolor{orange}{rgb}{1.0, 0.75, 0.0}
\begin{table}[htpb!]
    \caption{Quota countries: quota size, and quota baseline MMR level}
    \label{tab:qclass}
    \begin{tabular}{lcccc} \toprule
    Country & \multicolumn{2}{c}{Quota Size}  & \multicolumn{2}{c}{Baseline MMR} \\ \cmidrule(r){2-3} \cmidrule(r){4-5} 
    & Size  & Category                 & Value & Category \\ \midrule
    Afghanistan & 27.3 & \cellcolor{red}{High}    & 821   &  \cellcolor{red}{High}    \\
    Algeria     & 20 & \cellcolor{orange}{Medium} & 192   &  \cellcolor{green}{Low}   \\   
    Bangladesh  & 13 & \cellcolor{green}{Low}     & 481   &  \cellcolor{orange}{Medium}\\
    Burundi     & 30 & \cellcolor{red}{High}      & 993   &  \cellcolor{red}{High}     \\
    China       & 22 & \cellcolor{orange}{Medium} & 73    &  \cellcolor{green}{Low}   \\   
    Djibouti    & 10 & \cellcolor{green}{Low}     & 452   &  \cellcolor{orange}{Medium}\\
    Eritrea     & 30 & \cellcolor{red}{High}      & 931   &  \cellcolor{red}{High}    \\
    Haiti       & 3  & \cellcolor{green}{Low}     & 522   &  \cellcolor{orange}{Medium} \\
    Iraq        & 25 & \cellcolor{red}{High}      & 85    &  \cellcolor{green}{Low}   \\
    Jordan      & 11.1 & \cellcolor{green}{Low}     & 93    &  \cellcolor{green}{Low}   \\
    Kenya       & 13.4 & \cellcolor{green}{Low}     & 708   &  \cellcolor{red}{High}    \\
    Morocco     & 15.2 & \cellcolor{orange}{Medium} & 261   &  \cellcolor{green}{Low}   \\
    Niger       & 10 & \cellcolor{green}{Low}     & 825   &  \cellcolor{red}{High}    \\
    Pakistan    & 17.5 & \cellcolor{orange}{Medium} & 374   &  \cellcolor{green}{Low}   \\
    Rwanda      & 30 & \cellcolor{red}{High}      & 1220  &  \cellcolor{red}{High}    \\
    Saudi Arabia  & 20 & \cellcolor{orange}{Medium} & 20    &  \cellcolor{green}{Low}   \\
    South Sudan   & 25 & \cellcolor{red}{High}      & 857   &  \cellcolor{red}{High}    \\
    Sudan       & 25 & \cellcolor{red}{High}      & 626   &  \cellcolor{orange}{Medium} \\
    Swaziland   & 5.26 & \cellcolor{green}{Low} & 569   &  \cellcolor{orange}{Medium} \\
    Tanzania    & 29.1 & \cellcolor{red}{High}      & 946   &  \cellcolor{red}{High}    \\
    Uganda      & 24.4 & \cellcolor{orange}{Medium} & 673   &  \cellcolor{red}{High}    \\
    Zimbabwe    & 22.2 & \cellcolor{orange}{Medium} & 475   & \cellcolor{orange}{Medium} \\
    \bottomrule
    \end{tabular}
\end{table}
\clearpage

\begin{landscape}
\begin{table}[htpb!]
    \caption{Gender Quotas: TWFE impacts on women in parliament and maternal mortality}
    \label{difdif-full}
    \scalebox{0.92}{
    \begin{tabular}{lccccccccc}
    \toprule
    & \multicolumn{3}{c}{\% Women in Parliament} & \multicolumn{3}{c}{ln(Maternal Mortality Ratio)} & \multicolumn{3}{c}{Maternal Mortality Ratio} \\ \cmidrule(r){2-4} \cmidrule(r){5-7} \cmidrule(r){8-10}
    & (1) & (2) & (3) & (4) & (5) & (6) & (7) & (8) & (9) \\ \midrule
    \multicolumn{10}{l}{\textbf{Panel A: Intensity by Baseline MMR}} \\
    \input{./results/appendix/quotaDifDif_All_A.tex} \midrule 
    \multicolumn{10}{l}{\textbf{Panel B: Intensity by Quota Size}} \\
    \input{./results/appendix/quotaDifDif_All_B.tex}
    \bottomrule
    \multicolumn{10}{p{23.4cm}}{{\footnotesize Difference-in-differences (two-way fixed effect) estimates of the impact of reserved seats in parliament on women in parliament (columns 1-3), the log of the maternal mortality ratio (columns 4-6), and MMR in levels (columns 7-9) are displayed. In each case country and year fixed effects are included.  Baseline two-way fixed effect models are included in columns (1), (4) and (6), and then models studying heterogeneous impacts are presented there-after.  Heterogeneity is examined by baseline maternal mortality levels in panel A, and by the size of the gender quota in panel B.  In each case heterogeneity considers simple interactions between the reserved seat dummy and the intensity variable (baseline MMR or quota size respectively), and then separate intensity groups for low, medium and high intensity groups. Standard errors clustered by country are displayed in parentheses. * p$<$0.10; ** p$<$0.05; *** p$<$0.01.}}
    \end{tabular}}
\end{table}
\end{landscape}


\begin{table}
  \caption{Two-way fixed effect models: reserved seats and maternal mortality in India}
  \label{tab:India2wayFE}
  \begin{tabular}{lcccc} \toprule
    & \multicolumn{2}{c}{ln(MMR)}&\multicolumn{2}{c}{MMR} \\ \cmidrule(r){2-3}\cmidrule(r){4-5}
    & (1) & (2) & (3) & (4) \\ \midrule
    \input{./results/appendix/India/DD_India_baseline.tex}
    Population Weights & & Y & & Y \\
    \midrule
    \multicolumn{5}{p{12cm}}{\footnotesize Each column documents coefficients from two-way fixed effect models where the natural logarithm of the maternal mortality ratio (columns 1-2) or the maternal mortality ratio (columns 3-4) is regressed on an indicator for the state's reserved seat status, plus full state and time fixed effects.  Data covers the period from 1986 to 2003, from Sample Registration Systems and \citet{Rametal2006}.  Population weights based on census data are included in columns 2 and 4.  Standard errors clustered by state are presented in parentheses.    $^{***}p<0.01; ^{**}p<0.05; ^{*}p<0.10$.}  \\
    \bottomrule
    \end{tabular}
\end{table}

\begin{table}[htpb!]
    \caption{Weights and Estimates from the \citet{GoodmanBacon2018} decomposition}
    \label{tab:GBglobal}
    \begin{tabular}{lcc} \toprule
    & Weight & Estimate \\ \midrule
    \multicolumn{3}{l}{\bf Panel A: Women in Parliament} \\
    \input{./results/appendix/GB_womparl_all.tex} \midrule
    \multicolumn{3}{l}{\bf Panel B: ln(MMR)} \\
    \input{./results/appendix/GB_MMR_all.tex} \bottomrule
    \multicolumn{3}{p{8.6cm}}{\footnotesize Notes: The \citet{GoodmanBacon2018} decomposition above displays the weights and components making up the global ``single coefficient'' TWFE model. Decompositions are documented for the percent of women in parliament (panel A) and the natural logarithm of the MMR (panel B).  Each components' weight is given along with the point estimate for this comparison.  The global estimate is displayed at the foot of each panel.}
    \end{tabular}
\end{table}

\begin{table}[htbp]\centering
    \def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
    \caption{Reserved seats as an IV for women in parliament \label{qIV}}
    \begin{tabular}{l*{3}{c}}
    \toprule
    &(1) & (2) & (3) \\
    &\multicolumn{1}{c}{ln(MMR)}&\multicolumn{1}{c}{ln(MMR)}&\multicolumn{1}{c}{ln(MMR)}\\
    \midrule
    \multicolumn{4}{l}{\textbf{Panel A: LIML Estimates}} \\
    \input{./results/appendix/IVestimates_quota.tex}\midrule
    \multicolumn{4}{l}{\textbf{Panel B: First-Stage Estimates}}\\
    \input{./results/appendix/IVestimates_firststage.tex}
    \midrule
    \emph{Controls}: &&&\\
    Democracy \& growth & N & Y & N \\
    Empowerment \& predictors & N & N & Y \\   
    \bottomrule\multicolumn{4}{p{16.6cm}}{\begin{footnotesize} These are instrumental variables regressions in which implementation of gender quotas is used to instrument women in parliament.  The first stage regression of women in parliament on reserved seats is displayed in panel B of the table, while F-Statistics of the first stage and the associated p-value are presented below the principal estimates in Panel A. The first-stage F-stats are not consistently larger than standard rules of thumb. To avoid weak instrument problems, we use LIML rather than 2SLS as LIML has improved small sample properties when the IV is weak. Additionally, we present weak IV robust Anderson-Rubin confidence sets, which do not assume identification of coefficients. Power in this setting is graphed in Figure \ref{fig:weakIV}). The key take-away is that we can rule out values of zero or, in other words, our finding in the IV setting is robust to weak IV.     The displayed coefficients give the effect of an additional percentage of women in parliament on rates of maternal mortality, where women in parliament is instrumented with reserved seats.  The 90\% and 95\% confidence intervals from \citet{Conleyetal2012} are robustness tests, where we allow the instrument to be imperfect in the sense that the exclusion restriction holds approximately but not exactly, allowing quotas to have a direct effect in reducing MMR that is \emph{not} mediated by women in parliament of 0.01 (or 1\%). We use Conley et al. (2012)'s Union of Confidence Intervals (UCI) approach, with this process also based on LIML estimation. Each regression includes country and year fixed effects and clusters standard errors by country. Column 1 provides baseline models without controls beyond country and year FE.  In column 2 the regressions additionally control for time-varying (and potentially endogenous) GDP and democratization, while column 3 controls for an index of women's empowerment measures at baseline, interacted with year fixed effects. 
    * p$<$0.10; ** p$<$0.05; *** p$<$0.01.
    \end{footnotesize}}
    \end{tabular}
\end{table}

\begin{landscape}
    \input{./results/appendix/trendBreaks.tex}
\end{landscape}


\begin{landscape}
\begin{table}[htbp]
    \centering
    \caption{Alternative inference procedures for measures of maternal mortality in principal aggregate models}
    \label{tab:resample}
    \begin{tabular}{l*{6}{c}}
    \toprule
    &\multicolumn{2}{c}{\citeauthor{dCDH2020}}&\multicolumn{2}{c}{Two-way FE Estimator}&\multicolumn{2}{c}{\citeauthor{arkhangelsky2021synthetic}'s}\\
    &\multicolumn{2}{c}{$DID_M$ Estimator}&\multicolumn{2}{c}{}&\multicolumn{2}{c}{Synthetic DID Estimator}\\\cmidrule(lr){2-3}\cmidrule(lr){4-5}\cmidrule(lr){6-7}
    &\multicolumn{1}{c}{ln(MMR)}&\multicolumn{1}{c}{MMR}&\multicolumn{1}{c}{ln(MMR)}&\multicolumn{1}{c}{MMR}&\multicolumn{1}{c}{ln(MMR)}&\multicolumn{1}{c}{MMR}\\
    &(1) &(2) & (3) & (4) & (5) & (6)\\\midrule
    Reserved Seats (Point Estimate) & -0.072 & -86.46 & -0.082 & -106.10& -0.127 & -57.16\\
    &&&&&&\\
    p-value Bootstrap & 0.115 & 0.019 & 0.133 & 0.006 & 0.090 & 0.155\\
    p-value Triangular Correction & 0.161 & 0.027 & 0.260 & 0.047 & 0.209 & 0.317\\
    p-value Triangular Correction by Country & 0.124 & 0.014 & 0.149 & 0.004 & 0.099 & 0.163\\
    p-value Normal Correction & 0.301 & 0.035 & 0.617 & 0.120 & 0.539 & 0.427\\
    p-value Normal Correction by Country & 0.116 & 0.031 & 0.131 & 0.009 & 0.088 & 0.153\\\midrule
    Mean of Dep. Var.&       4.357&       233.425&       4.357&     233.425&     4.186&     182.757\\
    Observations     &        4,335&        4,335&        4,335&        4,335&        3,068&        3,068\\
    Number of Countries&      178&         178&         178&         178&         118&         118\\
    \bottomrule\multicolumn{7}{p{21.2cm}}{\begin{footnotesize} Notes:  Resample procedures are implemented following Appendix \ref{app:resample}.  These are based on single coefficient estimators from \citet{dCDH2020} (columns 1-2), two way FE estimates (columns 3-4) and \citet{arkhangelsky2021synthetic} (columns 5-6) based on the (required) balanced panel of observations.  Point estimates are presented, and below p-values associated with each point estimate, based on different procedures for re-sampling the uncertainty of measures of maternal mortality. In each case, re-samples are taken over country clusters, as treatment is defined at the level of the country.
    \end{footnotesize}}\end{tabular}
\end{table}
\end{landscape}


\begin{table}[htbp]
\centering
\caption{Mechanisms: impacts of gender quotas on intermediate outcomes\label{mDDmech}}
\scalebox{0.88}{
\begin{tabular}{l*{9}{c}}
\toprule
& Antenatal & Attended & Contraceptive & Fertility & Teenage   & Birth & Health & Development & GDP \\
& Care      & Births   & Usage         & Rate      & Pregnancy & Spacing & Expenditure & Assistance & per capita \\ 
&(1)& (2) & (3) & (4) & (5) & (6) & (7) & (8) & (9) \\ \midrule
\input{./results/appendix/quotaMechanism.tex}
\bottomrule\multicolumn{10}{p{21.6cm}}{\begin{footnotesize} Two-way FE models of intermediate outcomes as a function of the passage of gender quotas are displayed. Antenatal care coverage  and birth attendance are newly harmonized data available for 1990-2015 and measured as percentage coverage, however only available in a sub-sample of years for each particular country.  Contraceptive usage refers to the proportion of all women aged 15--49 using modern contraceptives.  Fertility rate is measured in natural logarithms, and teenage pregnancy is measured as births per 1,000 teenage women.  Birth spacing is measured in months to subsequent births, generated from full DHS data. Health expenditure refers to spending as a percentage of GDP, and development assistance refers to the proportion of development assistance directed to mothers health.  GDP per capita is measured in natural logarithms.  Standard errors clustered by country are displayed in parentheses. * p$<$0.10; ** p$<$0.05; *** p$<$0.01.
    \end{footnotesize}}\end{tabular}}
\end{table}




\begin{table}
    \centering
    \caption{Univariate regressions of ln(MMR) on mechanism Variables}
    \label{tab:univarMech}
    \begin{tabular}{lcc}
    \toprule
    Variable & Point & Standard \\
             & Estimate & Error \\ \midrule
    \input{results/appendix/simpleMechReg} \bottomrule
    \multicolumn{3}{p{7.6cm}}{\footnotesize Each line presents the point estimate and standard error from a projection of ln(MMR) on the variable indicated.  Standard errors are clustered by country, but no additional controls are included.}
    \end{tabular}
\end{table}
    
\begin{table}[htbp]
\centering
\caption{Gender quotas: TWFE impacts on alternative outcomes \label{difdif-other}}
\scalebox{0.9}{
\begin{tabular}{l*{9}{c}}
\toprule
&\multicolumn{3}{c}{Education}  &\multicolumn{3}{c}{Mortality}         &\multicolumn{3}{c}{Infant Mortality}       \\\cmidrule(lr){2-4}\cmidrule(lr){5-7}\cmidrule(lr){8-10}
& Female & Male & Female/Male & Tuberculosis & Female & Male & All & Female & Male \\
& Schooling & Schooling & Ratio & Mortality & Mortality & Mortality & IMR & IMR & IMR \\  
&\multicolumn{1}{c}{(1)}   &\multicolumn{1}{c}{(2)}   &\multicolumn{1}{c}{(3)}   &\multicolumn{1}{c}{(4)}   &\multicolumn{1}{c}{(5)}   &\multicolumn{1}{c}{(6)}   &\multicolumn{1}{c}{(7)}   &\multicolumn{1}{c}{(8)}   &\multicolumn{1}{c}{(9)}   \\
\midrule
\input{./results/appendix/quotaDifDif_otherOutcomes.tex}
\bottomrule\multicolumn{10}{p{20.8cm}}{\begin{footnotesize} Each column and panel presents estimates from a two-way FE specification following that in Table      \ref{difdif-full} (columns 1, 4 and 7).  We now examine estimates for impacts of gender quotas on alternative outcomes, specifically female and male education, as well as their ratio, and various classes of morbidity or mortality outcomes.  Educational data are drawn from Barro and Lee's education database, recording completed years of schooling of 15--19 year-olds in each country and are interpolated between quinquennial time period.   Morbidity/mortality outcomes are examined for adults (columns 4-6), and infants (columns 7-9). For adult morbidity/mortality, tuberculosis is measured as incidence per 100,000 people, and mortality is measured as per 1,000 adults of ages 15-60. Since the TB data had the occasional zero, we use the inverse hyperbolic sine transformation rather than the log transformation for this outcome, which is used for female and male mortality. Infant mortality is estimated from all publicly
available DHS waves and is expressed as the natural logarithm of infant mortality, for all births, and for males and females only. Standard errors clustered by country are displayed in parentheses. 
    * p$<$0.10; ** p$<$0.05; *** p$<$0.01.
    \end{footnotesize}}\end{tabular}}
\end{table}


\definecolor{LightGray}{rgb}{0.83, 0.83, 0.83}
\begin{table}[]
    \centering
    \caption{A back of the envelope calculation of impacts on total maternal deaths}
    \label{tab:boeMDeaths}
    \begin{tabular}{|ll|ll|} \hline
    \multicolumn{2}{|c|}{Baseline} & 
    \multicolumn{2}{c|}{\cellcolor{LightGray}{$\Delta MMR|Fertility$}} \\ \hline 
    Births = 34,735,750 & \multirow{2}{*}{MMR=267.5} &     
    Births = 34,735,750 & \multirow{2}{*}{MMR=244.2} \\
    Deaths = 92,928 &       &     Deaths = \textbf{84,843} &       \\ \hline
    \multicolumn{2}{|c|}{\cellcolor{LightGray}{$\Delta Fertility|MMR$}} & 
    \multicolumn{2}{c|}{\cellcolor{Gray}{$\Delta Fertility, \Delta MMR$}} \\ \hline   Births = 32,616,869 & \multirow{2}{*}{MMR=267.5} &     
    Births = 32,616,869 & \multirow{2}{*}{MMR=244.2} \\
    Deaths = \textbf{87,259} &       &     Deaths = \textbf{79,668} &       \\ \hline
    \multicolumn{4}{p{11.6cm}}{{\footnotesize The top left quadrant calculates maternal mortality ratios based on actual births and maternal deaths in all quota adopting countries one year prior to the quota.  The top right quadrant calculates total maternal deaths if the maternal mortality ratio declines by 8.7\%, but births remain fixed.  The bottom
    left quadrant calculates total maternal deaths if the fertility rate falls by 6.1\%, but the maternal mortality
    remains fixed at pre-quota levels.  Finally, the bottom right
    quadrant calculates the total reduction in expected maternal deaths given (a) the fall in fertility and (b) the fall in the maternal mortality ratio.}}
\end{tabular}
\end{table}


\begin{landscape}
  \input{./results/appendix/mechanisms_RHS.tex}
\end{landscape}

\begin{table}[htpb!]
\centering
\caption{Quota Adoption and Women's Inclusion Measures}
\label{tab:quotaAdop}
    \scalebox{0.9}{
    \begin{tabular}{lcccccc}\toprule
    & \multicolumn{2}{c}{\ \ \ \ \ \ Quota Adoption \ \  \ \ \ \ } & \multicolumn{2}{c}{\ \ \ \ Quota Adoption$_{(t+1)}$ \ \ \ \ } & \multicolumn{2}{c}{\ \ \ \ Quota Adoption$_{(t+2)}$ \ \ \ \ } \\ \cmidrule(r){2-3} \cmidrule(r){4-5} \cmidrule(r){6-7}
    & (1) & (2) & (3) & (4) & (5) & (6) \\ \midrule
    \input{./results/appendix/quotaAdoption.tex}
    \midrule
    Country and year FEs & Y & Y & Y & Y & Y & Y \\
    Transformed FEs &  & Y &  & Y &  & Y \\
    \bottomrule
    \multicolumn{7}{p{19.2cm}}{{\footnotesize Notes: Each specification regresses an indicator of a reserved seat quota being adopted at a certain time horizon (contemporaneously, or 1 or 2 years into the future) on a series of women's empowerment and political measures.  ``Transformed FEs'' refers to models which separate variables with considerable missingness out into fixed effects, including a separate FE for observations with missing information, so as to allow for greater data coverage.  These variables are the CIRI rights measures, propotion of women ministers, and measures of abortion legality.   Explanatory power refers to the within variation explained by these variables, while joint tests refers to the test of joint significance of all RHS measures. * p$<$0.10; ** p$<$0.05; *** p$<$0.01.}}
    \end{tabular}}
\end{table}

    








\begin{table}[htpb]
\centering
\caption{Two-way FE, DID$_M$, pooled event studies, and synthetic DID estimates with contemporaneous controls}
\label{tab:2wayFEcompCONTEMP}
\begin{tabular}{lcccccc} \toprule
& (1)     & (2)     & (3)     & (4)     & (5)     & (6)     \\ \cmidrule(r){2-7}
& \multicolumn{6}{c}{Outcome: ln(MMR)} \\ \midrule
%\input{./results/appendix/ModelComp_contemporaneous_lnMMRt1.tex}
\multicolumn{5}{l}{\textbf{Method A: Two-way FE Model}}\\ 
Reserved Seats&-0.082&-0.088*&-0.080&-0.070&-0.071&-0.062\\
    &(0.051)&(0.051)&(0.051)&(0.049)&(0.055)&(0.054)\\
\multicolumn{5}{l}{\textbf{Method B: $ DID\_M $ Estimates}}\\ 
Reserved Seats&-0.072*&-0.074*&-0.074*&-0.069&-0.080*&-0.083*\\
    &(0.043)&(0.043)&(0.044)&(0.044)&(0.047)&(0.050)\\
\multicolumn{5}{l}{\textbf{Method C: Pooled Event Study}}\\ 
Reserved Seats&-0.079**&-0.079**&-0.078**&-0.076**&-0.058&-0.050\\
    &(0.032)&(0.032)&(0.038)&(0.037)&(0.036)&(0.044)\\
\multicolumn{5}{l}{\textbf{Method D: Synthetic DID}}\\ 
Reserved Seats&-0.127**&-0.116*&-0.129*&-0.103&-0.128*&-0.099\\
    &(0.062)&(0.065)&(0.074)&(0.063)&(0.073)&(0.072)\\
\\ Negative Weights & -0.005 & -0.005 & -0.005 & -0.006 & -0.012 & -0.012 \\ 
   Observations & 4335  & 4335  & 4305  & 4324  & 4335  & 4294  \\ 
& \multicolumn{6}{c}{Outcome: Women in Parliament} \\ 
\midrule
%\input{./results/appendix/ModelComp_contemporaneous_womparl.tex}
\multicolumn{5}{l}{\textbf{Method A: Two-way FE Model}}\\ 
Reserved Seats&5.793***&5.628***&5.712**&5.747***&6.038***&5.660***\\
    &(2.167)&(2.027)&(2.227)&(2.159)&(2.145)&(2.052)\\
\multicolumn{5}{l}{\textbf{Method B: $ DID\_M $ Estimates}}\\ 
Reserved Seats&5.678**&5.172**&5.671**&5.674**&5.167**&4.644**\\
    &(2.222)&(2.021)&(2.304)&(2.226)&(2.154)&(2.018)\\
\multicolumn{5}{l}{\textbf{Method C: Pooled Event Study}}\\ 
Reserved Seats&6.622***&6.074***&6.546***&6.603***&6.242***&5.596***\\
    &(1.863)&(1.712)&(1.800)&(2.025)&(1.858)&(1.578)\\
\multicolumn{5}{l}{\textbf{Method D: Synthetic DID}}\\ 
Reserved Seats&8.281***&7.523***&8.361**&7.950**&7.661**&7.014**\\
    &(3.041)&(2.745)&(3.535)&(3.401)&(3.099)&(2.833)\\
\\ Negative Weights & -0.005 & -0.005 & -0.005 & -0.006 & -0.012 & -0.012 \\ 
   Observations & 4335  & 4335  & 4305  & 4324  & 4335  & 4294  \\ 
\midrule
\textbf{\small{Controls (contemporaneous)}}:&&&&&&\\
\small{Empowerment \& Predictors} && \small{Y} & & & & \small{Y} \\ 
\small{Democracy}   && & \small{Y} & & & \small{Y} \\ 
\small{Resources}   && & & \small{Y} & & \small{Y} \\
\small{Region$\times$year FE}    && & & & \small{Y} & \small{Y} \\
\bottomrule
\multicolumn{7}{p{15.6cm}}{\footnotesize Notes: Refer to notes to Table \ref{tab:2wayFEcomp}. Identical models are estimated, however in this case rather than including baseline$\times$post-quota controls, contemporaneous versions of all control variables are included. * p$<$0.10; ** p$<$0.05; *** p$<$0.01.}
\end{tabular}
\end{table}










\begin{landscape}
  \input{./results/appendix/rights_RHS.tex}
\end{landscape}
\begin{landscape}
  \input{./results/appendix/empower_RHS.tex}
\end{landscape}
\begin{landscape}
  \input{./results/appendix/polisci_RHS.tex}
\end{landscape}





\input{./results/appendix/quotaTimeImpacts_lnMMR.tex}
\input{./results/appendix/quotaTimeImpacts_womparl.tex}   

\begin{landscape}
\begin{table}[htpb!]
    \caption{Gender Quotas: TWFE impacts on women in parliament and maternal mortality (5 years pre/post adoption)}
    \label{difdif-full5y}
    \scalebox{0.92}{
    \begin{tabular}{lccccccccc}
    \toprule
    & \multicolumn{3}{c}{\% Women in Parliament} & \multicolumn{3}{c}{ln(Maternal Mortality Ratio)} & \multicolumn{3}{c}{Maternal Mortality Ratio} \\ \cmidrule(r){2-4} \cmidrule(r){5-7} \cmidrule(r){8-10}
    & (1) & (2) & (3) & (4) & (5) & (6) & (7) & (8) & (9) \\ \midrule
    \multicolumn{10}{l}{\textbf{Panel A: Intensity by Baseline MMR}} \\
    \input{./results/appendix/quotaDifDif_All_A_5years.tex} \midrule 
    \multicolumn{10}{l}{\textbf{Panel B: Intensity by Quota Size}} \\
    \input{./results/appendix/quotaDifDif_All_B_5years.tex}
    \bottomrule
    \multicolumn{10}{p{23.4cm}}{{\footnotesize Difference-in-differences (two-way fixed effect) estimates of the impact of reserved seats in parliament on women in parliament (columns 1-3), the log of the maternal mortality ratio (columns 4-6), and MMR in levels (columns 7-9) are displayed, following Table \ref{difdif-full}. However here quota adopting countries are only considered in the 5 years pre and post-quota adoption.  All other details follow those in Table \ref{difdif-full}. Standard errors clustered by country are displayed in parentheses. * p$<$0.10; ** p$<$0.05; *** p$<$0.01.}}
    \end{tabular}}
\end{table}
\end{landscape}

         
\begin{landscape}
  \input{./results/appendix/outlierAnalysis.tex}
\end{landscape}
    




%XXX to add robustness appendix somewhere i am putting it here for now

\clearpage  
\section{Appendix: Robustness Checks}
\label{app:Robustness}

We summarized robustness checks in the main text, here we provide full details.
\label{sscn:robustExtend}
\\
\noindent\textbf{Alternative estimators and pre-trends.} The placebo coefficients in the main results shown above are not significantly different from zero. The estimates stand up to including country-specific linear trends and region$\times$year fixed effects (Figure \ref{fig:robustness2alt}).
However, if standard tests of pretrends are underpowered, we might fail to capture the evolution of a relevant unobservable trend. To address this concern, we follow the ``Honest DiD'' procedure of \citet{RambachanRoth2019} and estimate upper and lower bounds on the dynamic effects %(corresponding to 95\% confidence intervals). 
For both women in parliament (Figure \ref{honest-womparl}) and MMR (Figure \ref{honest-mmr}) these bounds are informative at 95\% or, in some cases in later periods, at 90\%.\footnote{This exercise runs off the TWFE estimator. Below, we discuss that, in our setting, it delivers estimates similar to the preferred estimator.} 

Motivated by the possible concern that countries that do and do not adopt quotas are likely to be quite different, we generate synthetic controls for each treated country so as to achieve a match on pre-treatment levels of maternal mortality, drawing controls from the same geographic region as the treated country, and hence potentially subject to similar regional shocks. We aggregate estimates of treatment leads and lags across all treatment--synthetic control pairs, and implement an inference procedure based on clustered permutation to generate confidence intervals for these estimates. Appendix \ref{App:Match} details the procedures. The results in Figure \ref{fig:matchedControls} show no evidence of differential pre-trends, and a significant post-reform decline in MMR. We further re-estimate the models using the synthetic difference-in-differences procedure of \citet{arkhangelsky2021synthetic} and we observe, if anything, slightly larger effects, see panel D of Table \ref{tab:2wayFEcomp}. This procedure involves generating quota adoption year specific synthetic controls, placing more weights on countries and years which are more similar to treatment countries and periods.  A selection of year-specific treatment units and synthetic controls is presented in Appendix Figure \ref{fig:SDIDlnMMR}, along with details of their construction. These figures reveal gradual declines in MMR in treated units that are larger than in the respective synthetic controls.



Table \ref{tab:2wayFEcomp} shows that summary effect sizes from alternative estimators are fairly similar in magnitude. This suggests that, in our setting, the potential bias in the single coefficient two-way FE model discussed in \citet{GoodmanBacon2018} and \citet{dCDH2020} is in fact small. To illuminate this we provide the \citet{GoodmanBacon2018} decomposition of the identifying variation into its treatment vs.\ pure control and differential timing components (Table \ref{tab:GBglobal}). We find that 95.4\% of the estimated effect derives from the double-difference comparison of treated with never-treated units. The drop in MMR (of about 7\%) is similar when we compare early to late adopters (prior to adoption) to that obtained when comparing aggregate TWFE estimates of treated vs.\ never treated countries, albeit the weight attached to the latter is much greater (panel (b)).
The reason that results from two-way FE are close to those from the alternative estimators which more completely isolate the pure treatment versus control comparisons is that the majority of weights in the decomposition are attached to comparing treated to pure control units in a TWFE setting, with the share of units with `negative weights' \citep{dCDH2020} being small (Table \ref{tab:2wayFEcomp}).
Figure \ref{fig:GB} also reveals that the treatment versus pure-control estimates are quite closely clustered around the average effect (indicated by the dashed red line), which suggests that the observed reduction in MMR is observed broadly, rather than being driven by outliers.  Indeed, when we present leave-one-out estimates in Figures \ref{fig:DIDMLOO-MMR}--\ref{fig:LOO-WP}, these are indistinguishable from the main results, establishing that they are not driven by any particular country. 


So far we have displayed reduced form estimates. We also estimated LIML regressions of MMR on the share of women in parliament, instrumented with quota implementation.
If quotas were proxying an omitted variable the exclusion restriction would fail, but inference can still proceed on the premise of ``plausible exogeneity'', which delivers bounds on the IV estimates \citep{Conleyetal2012}. The estimates in Table \ref{qIV} provide the scaled impact of women's parliamentary representation among compliers. 
The IV point estimates indicate that a 1 percentage point increase in women's share in parliament is associated with a 1.5 to 2.0\% decrease in MMR. In estimating bounds, we allow the adoption of quotas to have a direct impact on MMR of up to $-$1\% over and above its impact on MMR via women in parliament. The estimated bounds are informative, indicating a 0.01\% to 3.7\% reduction in maternal mortality for a 1 percentage point increase in the share of women in parliament. We provide weak IV robust Anderson-Rubin confidence sets in Table \ref{qIV} and plot the implied rejection probabilities for the null across a range of null hypotheses in Figure \ref{fig:weakIV}, both of which suggest that the observed relationship between women in parliament and MMR is robust to corrections for weak instruments. These results add to the overall weight of the evidence provided in this section.

\noindent\textbf{Women's rights as predictors of gender quota legislation.} The multiple approaches to investigating pre-trends discussed in the preceding section allay the key identification concerns. We nevertheless directly investigate the possibility that our results derive from social preferences evolving gradually to favor gender equality, with gender quota legislation being one manifestation of this. This is plausible in view of historical evidence that laws are often passed once society is ready to comply with them \citep{DoepkeZilibotti2005,PLATTEAU2014633}. 
To measure social preferences and gender progressiveness in the policy environment, we pulled together data on 18 indicators of gender progressivity in the political, economic and civil domains including  indices of women's civil liberties, access to justice, economic rights, women's protests and the passage of abortion law (the full set of indicators is visible in the Figures and defined in the Data Appendix \ref{app:Data}). To examine the concern that the timing of quota adoption is a response to an upward drift in women's position, we use the same empirical strategy as for the main analysis, focusing on whether any of these 18 indicators shows an uptick prior to quota adoption. 
The estimates are in Figures \ref{fig:DIDMWomenRights2} and \ref{fig:DIDMWomenRights} (and the equivalent event study estimates are in Figures \ref{fig:eventWomenRights2} and \ref{fig:eventWomenRights}). 
The placebo coefficients of the \citeauthor{dCDH2020} estimates allow us to reject a positive pre-trend for each of the 18 indicators -- and thus to reject that quotas were adopted following improvements in other measures of equality for women. 
We bolster this evidence by regressing quota adoption on the full set of empowerment variables. We provide the F-statistic for joint significance of these variables, showing that it is small (Table \ref{tab:quotaAdop}). 

Our findings are plausible. While the progression of gender equality in society is likely to eventually culminate in increased attention to women's reproductive health, this is likely to be a slow process. In contrast, we discover that giving women instrumental power to directly influence policy can effect sharp change. We will pursue evidence of this in the section on mechanisms, where we use the same identification strategy to test for impacts of discontinuous changes in women's political power on a series of intermediate outcomes that (a) can be influenced by policy and (b) are known to bring MMR down. In contrast, as discussed above, few of the 18 indicators of women's equality are associated with MMR decline (Tables \ref{rightsRHS}--\ref{empowerRHS}).

One may be concerned that the findings reflect noisy measurement of the progressivity indicators. This concern is allayed in the specifications that involve putting them together on the right-hand side, or creating an index of them. It is further mitigated by evidence that the measures are meaningful insofar as we see improvements \emph{after} quotas are legislated in indices of women's political rights, political participation, power distribution by gender, and exclusion by gender. We discuss the post-quota (dynamic) coefficients in the Mechanisms section, as any changes that occur after quota adoption potentially reflect mechanisms by which quota adoption leads to MMR decline.



\noindent\textbf{Political variables as predictors of gender quota legislation.} If not women's rights then what is it that determines adoption of quota legislation? We could find no systematic quantitative analysis of this question, but, case studies in the political science literature indicate the possible relevance of pressure from international organizations (proxied by overseas development assistance), occasions of broader constitutional reform including regime transitions post-conflict reconstruction and the presence of peace-keeping forces \citep{Krook2009,BainesRubioMarin2005}.\footnote{Rwanda is a case in point. Women make up 62 percent of Rwanda's national legislature, the highest share in the world, and this happened as part of a major constitutional reform in the wake of the genocide. In Nepal there was similarly a major jump in the share of women in parliament in 2008 to 32.8\% in the wake of political transformation.} One might imagine that these same factors may have had direct impacts on MMR decline. To establish whether the estimated impacts of quota adoption might instead reflect  political changes, we 
follow the same approach as for women's rights, scrutinizing pre-trends in seven measures of the political environment, see Figure \ref{fig:dCDHPoliSci} (and see Figure \ref{fig:eventPoliSci} for the standard event studies). There are no significant placebo coefficients. The concern that our results are driven by political variables is further undermined by their bearing no clear association with MMR decline (Table \ref{polisciRHS}).



\noindent\textbf{Controlling for potential predictors of quota adoption.}  As discussed, we find no evidence that the 25 potential predictors, including women's rights and the political variables, predict quota adoption, or that they predict MMR. We nevertheless control in the main analysis for an index of the baseline (pre-quota, 1995) value of the indicators interacted with a dummy for post-quota years.\footnote{We use standardized versions of the underlying variables. We create separate indices, one for women's rights, one for the political variables and one for resources, which includes GDP, health expenditure per capita, and overseas development assistance for maternal health. We show that the results are not sensitive to whether we use these pre-quota indices times post-quota, or if instead we just include contemporaneous measures of the variables, see Table \ref{tab:2wayFEcompCONTEMP}.} 
We find that the estimated impact of quotas on MMR is larger and not statistically significantly different from the baseline estimate, see Figure \ref{fig:robustness2alt}, as well as full dynamic estimates with inference in Figures \ref{fig:DIDMReservedPred} and \ref{fig:eventReservedPred}. 
We also tested a series of placebo trend breaks, three to ten years before the date of quota legislation in each country, and find no evidence of a pre-legislation break in the trend in MMR (Table \ref{trendbreak}). 



\noindent\textbf{Endogenous changes in the composition of women giving birth.} 
If gender quotas lead to a shift in the composition of births such that women with lower baseline risks of maternal death are over-represented after quotas, then this could explain our finding of lower MMR. A compositional shift is not implausible given that we find an impact of quotas on fertility (Section \ref{scn:mechanisms}). We cannot investigate compositional change using aggregate country-year data on fertility, we need information on mother characteristics. To achieve this, we created a pseudo panel of births in the DHS data based on 10,837,442 births for 3,079,298 individual women from 82 countries surveyed in a total of 34 different years.\footnote{The DHS are available for 14 of 22 reforming countries and 59 of 156 non-reforming countries. Construction of the pseudo-panel on fertility is described in the supplementary data section, and total fertility rate (TFR) is constructed as a weighted average of age-specific fertility rates.} We model quota impacts on birth rates of women of different education and age categories, and on the sex ratio (male/female) of births, a proxy for maternal health \citep{Waldron1983,low2000sex}. We find no significant shifts in composition by any of these measures, see Figures \ref{fig:DHScharacDIDM}--\ref{fig:DHS}.
Our estimates are robust to controlling for time-varying measures of the age and educational composition of mothers, see Figure \ref{fig:robustness2alt} and Figure \ref{fig:eventAlternativesQuota} panel d).


\noindent\textbf{Measurement of maternal mortality.} Since MMR varies considerably across countries, proportional changes implied by using logarithms will exaggerate achievements in countries with lower baseline rates \citep{deaton2006global}, and we showed that treatment effects exhibit heterogeneity by baseline rates. We estimated an alternative model replacing the logarithm with the level of the MMR ratio, and the results hold (see Table \ref{difdif-full}).\footnote{The estimated impact is larger on account of outliers. If we winsorize MMR at 500 deaths per 100,000 live births, top-coding 806 observations with values above 500,  
the effect sizes in the levels model broadly agree with effect sizes in logs, see Table \ref{difdif-outliers} (where we additionally document other cutoffs, as well as dropping rather than winsorizing at the top end).}
A second issue is that the MMR data we use derive from vital statistics and demographic survey data, with gaps filled using modelled predictions \citep{Alkemaetal2015,Alkemaetal2016,WHO2015,Wilmothetal2012}. About 76\% of the country-year observations are original survey data points, the remaining 24\% being imputed. The data come with uncertainty intervals. We examine sensitivity of our estimates to this using three approaches. First, Figure \ref{fig:robustness2alt} shows that removing countries for which all observations are imputed has no substantive impact on the findings. Second, we directly account for this uncertainty, using a double-bootstrap procedure re-sampling over the uncertainty intervals to calculate the standard errors.
The results are in Table \ref{tab:resample}. Allowing for correlation within country reduces the estimated uncertainty. For the preferred \citet{dCDH2020} DID$_M$ estimator and the synthetic DID, with a triangular resampling procedure, the estimates tend to uphold the main results.
Third, we use an alternative measure of MMR which we derive from survey-based reports of sister deaths of DHS respondents, following the procedure detailed in \citet{BhalotraClarke2019}. The DHS maternal mortality module was implemented for 44 countries, of which 11 implemented quotas, and it covers the analysis period, 1990--2015. Using these data, we find a similar pattern of results (Figures \ref{DIDM-MMRDHS} and \ref{event-MMRDHS}). While less precisely estimated, the effects are larger, consistent with the DHS countries having higher MMR on average.

Another possible concern is that the availability and quality of MMR data is endogenous. If women parliamentarians, motivated by a concern to reduce MMR, improve surveillance and tracking of MMR (``we measure what we treasure"), this will render our estimates conservative. For instance, if women parliamentarians act to expand surveillance coverage by counting maternal deaths in remote under-developed areas then, other things equal, measured MMR will tend to increase. However, in principle this could go the other way if women politicians face more scrutiny of MMR and they react by pulling back on counting maternal deaths. Alternatively, if women politicians act to reduce the variance of (mean zero)
measurement error in MMR, this could make finding significant effects more likely. We acknowledge this problem without directly addressing it, but the fact that we identify mechanisms consistent with actual reductions in MMR suggests that our findings are less likely to be driven by changes in MMR measurement.



\noindent\textbf{Sensitivity to sample and clustering.} We explained earlier that including never-treated countries in the estimation sample aids identification of dynamic effects. We nevertheless assessed sensitivity to dropping the 51 high income countries and found that this yields essentially identical estimates (Figures  \ref{fig:eventAlternativesQuota}, \ref{fig:eventAlternativesQuotaWP}, \ref{fig:DIDMAlternativesQuota}, and  \ref{fig:DIDMAlternativesQuotaWP}), consistent with the MMR profile of these countries being relatively flat.\footnote{The sum of negative weights increases by 50\% when we remove high income countries, but it remains small.} When we further restrict the sample only to countries in Africa, we again find results consistent with those in the full sample (Figure \ref{fig:robustness2alt}), consistent with the majority of countries adopting quotas being African. To assess sensitivity to changes in the composition of countries in the panel, we dropped the 7 countries that passed quotas after 2005 to create a balanced sample with the baseline window of 10 years pre and post-quota. The estimates  are again unchanged (Figure \ref{fig:robustness2alt}). Estimates based on shorter time-horizons of 5 and 8 years pre- and post-quota adoption also agree with the baseline results using a 10 year window (Tables \ref{eventTimeMMR}, \ref{eventTimeWP} and 
\ref{difdif-full5y}). 

In Section \ref{scn:mechanisms} we will analyse mechanisms. For some mechanisms variables, the data are sparser than for the main results. We therefore re-estimate the main results on the common sample, see Figure \ref{fig:eventMechanismBAL} (and Figure \ref{fig:DIDMMechanismBAL} for corresponding DID$_{M}$ plot). The results are less precise in this smaller sample but the broad patterns remain. 
Inference in our main specifications treats the data as independent across countries, but not within countries. To address the potential concern that quota implementation was temporally correlated, we estimate event studies with two-way clustering \citep{Cameronetal2011} of standard errors by both country and by year, see Figures \ref{fig:eventAlternativesQuota}--\ref{fig:eventAlternativesQuotaWP}. While the confidence intervals are now wider, we still observe statistically significant effects. 



     
\clearpage  
\section{Data Appendix}
\label{app:Data}
The following table provides an exhaustive list of the variables used in the paper along with the original sources. 

\begin{scriptsize}
\begin{longtable}{p{.35\textwidth} p{.60\textwidth}} 
\hline
\textbf{Variable (Source)} & \textbf{Description} \\ \hline
\endhead
\hline
\multicolumn{2}{r}{\scriptsize (continued on next page)} \\
\endfoot
\endlastfoot
\textit{Maternal Mortality, Cross Country} (MMEIG -- modelled estimates) &  We used recently released estimates of the maternal mortality ratio (MMR) per 100,000 live births produced by the Maternal Mortality Estimation Inter-Agency Group (MMEIG) and published in the World Bank World Development Indicators (WDI, indicator SH.STA.MMRT). These data were made available for the first time in the year 2016 and before that there were no reliable annual cross-country data on MMR.  These estimates were available for 183 countries annually for the period 1990--2015.  Maternal mortality is identified using ICD-10 codes O00-O99 (Pregnancy, childbirth and puerperium); the official definition is ``the number of women who die from pregnancy-related causes while pregnant or within 42 days of pregnancy termination per 100,000 live births.'' These are widely considered the best MMR measures to date, as they address known measurement difficulties in survey and vital statistics data on maternal mortality using Bayesian methods applied to multiple, complementary data sources including vital statistics, special inquiries, surveillance sites, population-based household surveys and census files \citep{Alkemaetal2016,Alkemaetal2015}. The world distribution of average MMR for the period of 1990--2015 is in Figure \ref{MMRworld}. \\ [0.8ex]
\textit{Maternal Mortality, Indian states } (SRS) &  State-level estimates of maternal mortality in India are gleaned from the Office of the Registrar General \& Census Commissioner’s Sample Registration System (SRS) Bulletins \citep{general2006sample}. These data are available for the 15 large states -- Andhra Pradesh, Assam, Bihar (including Jharkhand), Gujarat, Haryana, Karnataka, Kerala, Madhya Pradesh (including Chattisgarh), Maharashtra, Odisha, Punjab, Rajasthan, Tamil Nadu, Uttar Pradesh (including Uttarakhand), and West Bengal -- for the period of 1997--2003. These are complemented by an aggregate (state-level) estimate from 1986, reported in \citet{Rametal2006} covering the same large states.\\ [0.8ex]    
\textit{Ethnic Compostion} (HIEF) & Ethnic composition is measured by Ethnic Fractionalization (EF Index), which is the probability that two individuals who are chosen at random from a country belong to two different ethnic groups. Time-varying data on EF come from Historical Index of Ethnic Fractionalization Dataset (HIEF) V 2.0 \citep{dravzanova2020introducing}, downloaded from the Harvard Dataverse (\url{https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/4JQRCL}). The dataset is available for the years 1945--2013.\\ [0.8ex]
\textit{Political Gender Quota Data, Cross-Country} (IDEA) &  We collated measures for each
    country of whether the country has a legislated and binding reserved
    seat quota for women, its year of implementation, and the size of the
    quota measured as number of seats divided by all seats in the uni- or
    bi-cameral chamber.  To create the database, we started with measures
    provided by \citet{Dahlerup2005} and completed the most recent years
    from Global Database of Quotas for Women database  
    which is a repository developed and maintained by
    the International Institute for Democracy and Electoral Assistance (IDEA),
    the Inter-Parliamentary Union, and Stockholm University.% (available online at \url{https://www.idea.int/}). 
    %This database was consulted on 19th of July, 2016
		\\ [0.3ex]	
\textit{Political Gender Quota Data, India} (\cite{Iyeretal2012}) & State-level data on the reservation of seats for women in local  councils come from \cite{Iyeretal2012} who provide a time-varying state-level indicator for when women were given political representation. ``[T]he indicator equals one in the years following the first local government election that implemented
the “not less than one-third” reservation scheme for women representatives" \citep{Iyeretal2012} .\\ [0.3ex]			
\textit{Women in Parliament Data} (WDI, MDG, \citep{Paxtonetal2008}) & We used three distinct annual-level
    measures of women in parliament to construct a comprehensive
    panel of the percentage of women occupying seats in the national
    parliament.  These were the WDI indicator SG.GEN.PARL.ZS
    (``Proportion of seats held by women in national parliaments (\%)''),
    The UN Millennium Development Goals (MDG) Indicators (``Seats held by
    women in national parliament, percentage''), and the Inter university
    Consortium for Political and Social Research (ICPSR) dataset compiled
    by \citep{Paxtonetal2008} (``Women in Parliament, 1945--2003:
    Cross-National Dataset''). The first two of these datasets had partially-complete coverage for the
    years 1990, and then 1997--2015, while the latter had partially-complete
    yearly coverage for each year starting in 1945, and ending in 2003. In
    order to construct as comprehensive a series as possible, we began with
    the WDI data, and then imputed missing years where available from the MDG
    indicators, and \citet{Paxtonetal2008} data.  When a missing WDI year
    was available in both the MDG and the ICPSR dataset, we favored the MDG
    measure, which was estimated using the same sample and year.
    Figures \ref{womparlProp} and \ref{womparlPropAll} present the distribution of the proportion of women in parliament pre- and post-quota implementation in quota countries, as well as the full distribution of the proportion of women in parliament over the period under study. \\ [0.8ex]
\textit{Health Expenditure} (WHO) &  Health expenditure at the country-year level was taken from the World Health Organization the National Health Accounts (NHA) data series.  These provide a measure of total health expenditure as a percent of GDP, and are available for the years 1995--2013.  %For countries in which data exists for the majority of the analysis period, an identical multiple imputation procedure to that previously described is followed to extrapolate expenditure data to the years 1990-1994 and 2014-2015. Information on variable creation can be found here: https://apps.who.int/nha/database/DocumentationCentre/Index/en
\\ [0.8ex]
\textit{Development assistance for health} (IHME) & Development assistance for health (DAH) data are based on the Institute for Health Metrics and Evaluation (IHME) Development Assistance for Health Database (1990--2017). These data are available at the source country $\times$ receiver country $\times$ year level. We compute the proportion Development Assistance for Health to Maternal Health as: $\frac{\text{Development Assistance for Health to Maternal Health - All Program Areas (constant 2017 US dollars)}}{\text{Development Assistance for Health  disbursed from all channels (constant 2017 US dollars)}}$. 
\\ [0.8ex]
\textit{Women's Protests} \citep{bell2019impact} & We compute the proportion of women's protests using country-level panel data on women's protests and total protests from \cite{bell2019impact}. (We are grateful to Sam Bell for sharing the data with us.)  
\\ [0.8ex]
\textit{Contraception Prevalence} (UN -- modelled estimates) & This variable measures the percentage of Women of reproductive age (15-49 years) who are currently using any modern method of contraception. The annual data are from the ``ESTIMATES AND PROJECTIONS OF FAMILY PLANNING INDICATORS 2020" dataset generated by the UN Department of Economics and Social Affairs, Population Division. 
Downloaded from \url{https://www.un.org/development/desa/pd/data/family-planning-indicators}  on 03/02/2021. \\ [0.8ex]
\textit{Adult Male \& Female Mortality} (World Bank) & Data on male/female mortality for adults are available in the World Bank Data Bank (indicators SP.DYN.AMRT.MA and SP.DYN.AMRT.FE), based on measures from the United Nations Population Division, World Population Prospect and University of California, Berkeley, and Max Planck Institute for Demographic Research. This is measured as mortality between the ages of 15--60, per 1,000 male adults, and captures the likelihood that a male/female of age 15 dies by the age of 60. 
\\[0.8ex]
\textit{Tuberculosis Mortality} (WHO) & This is measured as the number of deaths due to Tuberculosis among HIV negative people, and is measured per 100,000 population. The data are from the WHO and were downloaded from: \url{http://apps.who.int/gho/data/view.main.57020ALL?lang=en}, accessed on 17/03/2016. Further details on estimation are provided here: \url{https://www.who.int/data/gho/indicator-metadata-registry/imr-details/1425}\\ [0.8ex]
\textit{Women's Rights} (CIRI) &  Measures of women's political, economic, and social rights data are based on the \citet{CingranelliData2013} data set available from \url{http://www.humanrightsdata.com/}. We use the three following variables:\\ [0.8ex]
&- \textit{Women’s Political Rights}: takes into account women's rights to vote, to run for political office, to hold elected and appointed government positions, to join
political parties, and to petition government officials. This variable takes discrete values between 0 and 3, with 3 representing high rights and 0 representing low rights. This variable is available for the period of 1981--2011 for approximately 139 (in 1990) to 195 (in 2011) countries.\\ [0.3ex]	
&- \textit{Women’s Economic Rights}: takes into account women's rights to equal pay for equal work, free choice of profession or employment without the need to obtain a husband or male relative's consent, the right to gainful employment without the need to obtain a husband or male relative's consent, equality in hiring and promotion practices, job security (maternity leave, unemployment benefits, no arbitrary firing or layoffs, etc.), non-discrimination by employers, be free from sexual harassment in the workplace, to work at night, to work in occupations classified as dangerous, to work in the military and the police force. This variable is available for the period of 1981--2011 for approximately 139 (in 1990) to 195 (in 2011) countries.\\ [0.3ex]	
&- \textit{Women's Social Rights}: takes into account women's rights to equal inheritance; enter into marriage on a basis of equality with men; travel abroad;  obtain a passport; confer citizenship to children or a husband;  initiate a divorce; own, acquire, manage, and retain property brought into marriage;  participate in social, cultural, and community activities;  education; and freedoms to choose a residence/domicile, and from female genital mutilation of children and of adults without their consent, and from forced sterilization. This variable is available for the period of 1981--2007 for approximately 139 (in 1990) to 193 (in 2007) countries.\\ [0.3ex]	
\textit{Women's Rights} (VDEM) &  Measures of women's rights and measures of relative social standing are based on data collected in the Variety of Democracy dataset  Version 10 \citep{VDEM2020,mm2020} -- VDEM-10 --  downloaded from \url{https://www.v-dem.net/en/data/data/v-dem-dataset/} on the  27/10/2020. When generating ratios of women to male outcomes these
        are Winsorized at the 1\textsuperscript{st} and 99\textsuperscript{th}
        percentiles to avoid outliers in cases where the denominator is very
        small.  We follow suggested practices in removing a small number of
        observations which are coded by 3 or fewer country experts \citep{VDEM2020}
        in ``C'' type variables from the VDEM data which are based on the
        opinions of country experts. We specifically use the following variables: \\ [0.8ex]
&- \textit{Women’s Civil Liberties Index ($v2x\_gencl$ -- modelled estimates)}: ``The index is formed by taking the point estimates from a Bayesian factor analysis model of the indicators for freedom of domestic movement for women ($v2cldmovew$), freedom from forced labor for women ($v2clslavef$), property rights for women ($v2clprptyw$), and access to justice for women ($v2clacjstw$)" \citep{Vdemcodebook2020}. \\ [0.3ex]	
&- \textit{Women’s Political Participation Index ($v2x\_genpp$)}: ``The index is formed by taking the average of the indicators for lower chamber female legislators ($v2lgfemleg$, standardized) and power distributed by gender ($v2pepwrgen$)" \citep{Vdemcodebook2020}.\\ [0.3ex]	
&- \textit{Exclusion by Gender Index ($v2xpe\_exlgender$ -- modelled estimates)}: ``The index is formed by taking the point estimates from a Bayesian factor analysis model of the indicators power distributed by gender ($v2pepwgen$), equality in respect for civil liberties by gender ($v2clgencl$), access to public services by gender ($v2peapsgen$), access to state
jobs by gender ($v2peasjgen$), and access to state business opportunities by gender ($v2peasbgen$)" \citep{Vdemcodebook2020}.\\ [0.3ex]	
&- \textit{Relative Freedom of Movement for Women}: Ratio of $\frac{\text{freedom of domestic movement for women (v2cldmovew)}}{\text{freedom of domestic movement for women (v2cldmovem)}}$. The indicators specify ``the extent to which all women / men are able to move freely, in daytime
and nighttime, in public thoroughfares, across regions within a country, and to establish permanent residency where they wish" \citep{Vdemcodebook2020}.  \\ [0.3ex]	
&- \textit{Relative Access to Justice for Women}: Ratio of $\frac{\text{Access to justice for women (v2clacjstw)}}{\text{Access to justice for men (v2clacjstm)}}$. Access to justice ``specifies the extent to which women/men can bring cases before the courts without risk to their personal safety, trials are fair, and women have effective ability to seek redress if public authorities violate their rights, including the rights to counsel, defense, and appeal" \citep{Vdemcodebook2020}.\\ [0.3ex]	
&- \textit{Relative Freedom of Discussion for Women}: Ratio of $\frac{\text{Freedom of discussion for women (v2cldiscw)}}{\text{Freedom of discussion for men (C) (v2cldiscm)}}$. Freedom of discussion ``specifies the extent to which [women/]men are able to engage in private discussions,
particularly on political issues, in private homes and public spaces (restaurants, public transportation, sports events, work etc.) without fear of harassment by other members of the
polity or the public authorities. We are interested in restrictions by the government and its agents but also cultural restrictions or customary laws that are enforced by other members of the polity, sometimes in informal ways" \citep{Vdemcodebook2020}.\\ [0.3ex]	
&- \textit{Power distributed by gender ($v2pepwrgen$)}: Measures if political power distributed according to gender. Lower values indicate men have near-monopoly on political power, while higher values indicate more equality in power between women and men. \\ [0.3ex]	
&- \textit{Freedom from forced labor for women($v2clslavef$)}: measures whether adult women free from servitude and other kinds of forced labor. ``Involuntary servitude occurs when an adult is unable to quit a job s/he desires to leave — not by reason of economic necessity but rather by reason of employer’s coercion. This includes labor camps but not work or service which forms part of normal civic obligations such
as conscription or employment in command economies" \citep{Vdemcodebook2020}.\\ [0.3ex]
&- \textit{Property rights for women ($v2clprptyw$)}: measures whether women enjoy the right to private property. ``Private property includes the right to acquire, possess, inherit, and sell private property, including land. Limits on property rights may come from the state (which may legally limit rights or fail to enforce them); customary laws and practices; or religious or social norms. This question concerns the right to private property, not actual ownership of property." \citep{Vdemcodebook2020}.\\ [0.3ex]
\textit{Women, Business and the Law Index} (World Bank) & The WBL index is a recently released index covering 170 countries between 1970--2020, providing a measures of women's equality in development outcomes, labor force participation, vulnerable employment and political participation where scores are based on the average of each economy’s scores for 8 distinct topics listed below. A higher score indicates more gender equal laws.
\\ [0.8ex]
&Mobility -- Incorporates answers to the following questions -- Can a woman apply for a passport in the same way as a man? Can a woman travel outside the country in the same way as a man? Can a woman travel outside her home in the same way as a man? Can a woman choose where to live in the same way as a man?\\ [0.3ex]
&Workplace -- Incorporates answers to the following questions -- Can a woman get a job in the same way as a man? Does the law prohibit discrimination in employment based on gender? Is there legislation on sexual harassment in employment? Are there criminal penalties or civil remedies for sexual harassment in employment?\\ [0.3ex]
&Pay -- Incorporates answers to the following questions -- Does the law mandate equal remuneration for work of equal value? Can a woman work at night in the same way as a man? Can a woman work in a job deemed dangerous in the same way as a man? Can a woman work in an industrial job in the same way as a man? \\ [0.3ex]
&Marriage -- Incorporates answers to the following questions -- Is there no legal provision that requires a married woman to obey her husband? Can a woman be head of household in the same way as a man? Is there legislation specifically addressing domestic violence? Can a woman obtain a judgment of divorce in the same way as a man? Does a woman have the same rights to remarry as a man?\\ [0.3ex]
&Parenthood -- Incorporates answers to the following questions -- Is paid leave of at least 14 weeks available to mothers? Length of paid maternity leave? Does the government administer 100\% of maternity leave benefits? Is there paid leave available to fathers? Length of paid paternity leave? Is there paid parental leave? Shared days? Days for the mother? Days for the father? Is dismissal of pregnant workers prohibited?\\ [0.3ex]
&Entrepreneurship -- Incorporates answers to the following questions -- Can a woman sign a contract in the same way as a man?	Can a woman register a business in the same way as a man?	Can a woman open a bank account in the same way as a man?	Does the law prohibit discrimination in access to credit based on gender?
 \\ [0.3ex]
&Assets -- Incorporates answers to the following questions -- Do men and women have equal ownership rights to immovable property?	Do sons and daughters have equal rights to inherit assets from their parents?	Do male and female surviving spouses have equal rights to inherit assets?	Does the law grant spouses equal administrative authority over assets during marriage?	Does the law provide for the valuation of non-monetary contributions?\\ [0.3ex]
&Pension -- Incorporates answers to the following questions -- Is the age at which men and women can retire with full pension benefits the same? Is the age at which men and women can retire with partial pension benefits the same? Is the mandatory retirement age for men and women the same? Are periods of absence due to childcare accounted for in pension benefits?\\ [0.3ex]
\textit{Women's Rights} (additional data) & We gleaned Women’s rights data from several additional sources:\\ [0.8ex]  
&- \textit{Women Ministers}: Women Minister data is drawn from the Inter Parliamentary
Union (IPU).  These were compiled based on country year aggregates for 1994, 1998, 2000, 2005, 2008, 2010, 2012, 2014 and 2015.\\ [0.3ex]	
&- \textit{Female Labor Force Participation -- modelled estimates}: This is modelled by the International Labour Organization (provided in the ILOSTAT database) as the percent of women aged 15 and above who are economically active.  It is available from 1990 to 2015.\\ [0.3ex]
&- \textit{Abortion (Save Mother's Life)}: Measures of abortion availability, and in which circumstances abortion is legal is drawn from the data of \citet{Eliasetal2017}.\\ [0.3ex]
&- \textit{Abortion (Fetal Impairment)}: As above. \\ [0.3ex]
$^{\ddag}$\textit{Maternal Care Inputs} (World Bank) & Recent data from the World Bank Data Bank allow us to examine the state of maternal health care in a subset of countries and years. These data are constructed and released by the World Bank using comparable measures from each country: specifically data from UNICEF, the State of the World’s Children, Child Info, and the Demographic and Health Surveys (DHS). As such, these measures are only available in years and countries for which surveys were conducted, resulting in fewer observations than the yearly measures of maternal mortality. In our analysis we use the full set of data released in the World Bank Data Bank. We use the following two policy relevant indicators: \\ [0.8ex]
&- \textit{Antenatal Care}: The percent of pregnant women receiving prenatal care (indicator SH.STA.ANVC.ZS). \\ [0.3ex]	
&- \textit{Skilled Birth Attendance}: the percent of all births attended by skilled health staff (indicator SH.STA.BRTC.ZS). \\ [0.3ex]	
\textit{Fertility} (World Bank) & Total fertility rates are expressed as the number of children expected to be born per woman based on current fertility rates, and are available as World Bank indicator SP.DYN.TFRT.IN.\\ [0.8ex]
$^{\ddag}$\textit{Women's Schooling, Average 15--19} \citep{BarroLee2012}& Women’s schooling measures are taken from the Barro-Lee dataset \citep{BarroLee2012} which gives average years of schooling for women aged 15--19 years. Barro-Lee is only available quin-quennially from 1950 to 2015. We use the sample from 1990–2015, and linearly interpolate by country between 5 year periods.\\ [0.8ex]
$^{\ddag}$\textit{Men's Schooling, Average 15--19} \citep{BarroLee2012}& As above.\\ [0.8ex]
\textit{Teen Pregnancy}  (World Bank) &  The adolescent fertility rate expressed as the number of births per 1,000 women aged 15--19 is provided by the World Bank, WDI database, indicator SP.ADO.TFRT.\\ [0.8ex]
\textit{GDP} (World Bank) & The log of GDP per capita is PPP adjusted and measured in 2011 international dollars. This is taken from the World Bank WDI database, indicator NY.GDP.PCAP.PP.KD.\\ [0.8ex]
\textit{Infant Mortality Rates} (DHS) & We use male and female infant mortality rates based on the DHS data. Infant mortality is defined as the death of a child before reaching the age one.\\ [0.8ex]
\textit{Quota Predictors} (IPI, World Bank, \cite{Becketal2001} ) & We examined quota predictors as laid out in \citet{Krook2009}:\\   
&- \textit{Peacekeepers}: the number of peacekeepers in a
    country from The International Peace Institute, IPI Peacekeeping Database. \\ [0.3ex]
&- \textit{Net Overseas Development Assistance}: World Bank Indicator DT.ODA.ODAT.CD \\ [0.3ex]    
&- \textit{Political Competition}: a series of measures of political competition and landscape from
    \citep{Becketal2001} \\ [0.3ex]    
  
\textit{Democracy} \citep{boix2013complete} & Our dichotomous measure of democracy was gleaned from the \cite{boix2013complete} database (updated in 2018). This database provides a binary index of democracy, measured annually, for all sovereign countries (222 distinct countries), between 1800 and 2015.  They classify a country at a given time as a democracy if it meets two criteria: high political contestation (the decisions to govern the state are taken through voting procedures that are free and fair), and high participation (a minimal level of suffrage).\\ [0.8ex]
\textit{Population} (UN) & We collect population covering the entire population, and just the population of women aged 15--64 years which are available by country and year over an extended period of time.  These are drawn from the United Nations Population Division's World Population Prospects. \\[0.8ex]
\textit{Maternal Mortality} (DHS) & We construct country by year measures of maternal mortality based on microdata collected in the DHS.  The DHS employs the sisterhood method, asking each woman to list all her sisters, whether they are surviving, and if not, the year of their death and whether this was related to maternal causes.  The DHS maternal mortality module which employs this procedure is implemented in only a sub-sample of countries.  We collect all DHS survey waves which have implemented the maternal mortality module, and use these to construct retrospective panels of all sisters listed by women, converted into pseudo-panels such that each sister has an observation for each year in which she is aged between 15--49 and is surviving.  For each sister$\times$year observation we infer whether she dies for reasons related to childbirth based on the sisterhood surveys.  Based on this pseudo-panel, we calculate a country by year observation for rates of maternal mortality. \\[0.8ex]
\textit{Fertility, birth spacing and birth composition} (DHS) & Country by year measures of age-specific fertility rates, total fertility, birth spacing and characteristics of mothers giving birth are calculated from DHS microdata based on retrospective fertility panels.  Each surveyed woman is asked to report her full fertility history including months and years of all births.  From these fertility histories we generate a pseudo-panel covering all women with a single observation for each of their ages between 15--49 (or up to their age at the time of the survey).  In each year, we infer whether the woman gave birth, and if so, the birth spacing between that and the following birth, as well as the child's sex.  Based on these pseudo-panels we aggregate data to country$\times$year cells calculating fertility rates in each quinquennial age group (15--19, 20--24, \ldots, 40--44), as well as the average birth spacing for all births, the sex ratio of births, and the mean education and literacy status of mothers giving birth.    \\[0.8ex]
\textit{Infant Mortality by sex} (DHS) & We calculate (total) infant mortality as well as infant mortality by sex from DHS microdata.  To do so, we generate a panel of all children born reported by mothers in the DHS, their year of birth, their gender, and their survival status.  If they are recorded as not surviving, we record their age at death.  From these micro-level observations, we calculate the infant mortality rate as the total number of infant deaths in the country and year cell, divided by the total number of births.  A similar calculation is conducted for male and female infant mortality rates (with the denominator of male and female births respectively).  \\[0.8ex]
\hline
\multicolumn{2}{p{.95\textwidth}}{\scriptsize \emph{Notes.} $^{\ddag}$ For indicated time-varying variables, missing values are linearly interpolated. Where relevant, tests using original (non-interpolated) values are provided as part of the online appendix.  }
\end{longtable}
\end{scriptsize}

\clearpage
    \section{Matched Controls using a Pooled Synthetic Control Procedure}
\label{App:Match}
In the main event study specification identification is drawn by comparing changes in countries which adopt quotas to all other countries which did not adopt quotas.  Thus, `control' units consist of all non-quota countries, with prevailing differences (in the pre-reform period) captured by country fixed effects. An alternative model consists of explicitly matching each treated country with a single aggregate or synthetic matched control country, and observing how outcomes evolve in the treated and synthetic control, inferring treatment effects based on differences in each case.

We conduct such a procedure, generating a synthetic control for each quota country.  We then generate a full dynamic set of treatment effects by pooling lags across each post-quota period as laid out below.  Consider country $c$ which adopted quotas in year $t$.  For this country, the pool of countries which can potentially generate a synthetic control consists of all 156 countries which never adopted a reserved seat quota.  For country $c$ generate the synthetic control following \citet{Abadieetal2010} by finding the vector of weights $\mathbf{W}^*$ which minimize: $||\mathbf{X_1}-\mathbf{WX_0}||$
where $\mathbf{X_1}$ and $\mathbf{X_0}$ consist of a matrix of (identically defined) matching variables between the quota country and the pool of non-quota countries.  In this particular implementation, the matching variables consist of rates of the outcome of interest (women in parliament or maternal mortality) in each of the pre-quota periods $t-10, t-9, \ldots, t-3$, as well as binary indicators for the country's sub-region based on UN sub-region classifications.  Note that there are three particular features of this method that should be highlighted.  The first is that countries will be explicitly matched based on pre-treatment outcomes such that pre-trends in country $c$ and its synthetic control should be equal in levels (this is a test of the quality of the synthetic control).  Second is that periods $t-2$ and $t-1$ are not mechanically matched in the synthetic control procedure, and thus can act as placebo periods such that impacts should not be observed prior to the reform's passage.  And finally, note that the inclusion of sub-region binary indicators is a device which favors choosing matched units from within the same geographic areas (and potentially exposed to similar shocks), but that units from outside of these areas can be included if the difference in binary variables owing to sub-region variations is offset by the weighted quadratic difference in other matching variables.\footnote{In other words, countries from outside of the sub-region can be included in the synthetic control if they are considerably better matches than countries within the same sub-region and can pay the penalty based on different sub-region indicators in the matching process.}

We conduct this procedure for each quota country which is amenable to generation of a synthetic control in this way.  Note that of the 22 quota countries, only 15 can be adequately matched in synthetic control methods.  Uganda adopts quotas prior to 1990 and so has no pre-adoption periods to match on.  South Sudan is a new country, and so has insufficient periods to match on.  Afghanistan, Burundi, Iraq, Djibouti and Saudi Arabia each have multiple missing measures for MMR in the years prior to quota adoption and as such have missing entries in the $\mathbf{X_1}$ matrix.  Once each country has its own synthetic control generated, we estimate full dynamic treatment effects by averaging the difference between each post-treatment period and the synthetic control over all treatment-synthetic control pairs.  Thus, the estimated treatment effect on the first lag is the mean of the difference between each country and its synthetic control in the first post-quota lag, with similar procedures conducted for each lead and lag term.

Finally, to conduct inference in this setting we implement a permutation method, which consists of randomly permuting the actual distribution of quota years across alternative countries. Thus, each permutation maintains the precise structure of the data, simply permuting treatment units.  We conduct 500 permutations, and generate 95\% confidence intervals for \emph{each} lag and lead as the end points of percentile 2.5 and 97.5 of each permuted treatment effect.


\section{Description of Resampling Procedure for MMR Uncertainty}
\label{app:resample}
Publicly available MMR data published by the MMEIG consist of a point estimate and the upper and lower points of the 80\% uncertainty interval. In describing the modeling procedure, the authors note ``We computed 80\% uncertainty intervals (UIs) for the MMR and all related outcomes using the 10th and 90th percentiles of the posterior distributions. \ldots  We report 80\% UIs rather than 95\% UIs because of the substantial uncertainty inherent in maternal mortality outcomes: intervals based on higher uncertainty levels quickly lose their ability to present  meaningful summaries of a range of likely outcomes.'' 	\citep[p. 1250]{Alkemaetal2016}. In order to estimate standard errors and p-values based on these data, we undertake the following procedure:
    
    
       \textbf{Resample Algorithm}
       \begin{enumerate}
       \item Take a clustered bootstrap resample from the original data $b=1,\ldots,B$, with $B=50$
       \item Generate a random vector of size $N$ (where $N$ is the sample in the regression) where
         each element is either a) a draw of a normal variable where 80\% of the probability mass falls
         between -1 and 1 (a draw from $\mathcal{N}(0,0.7803)$), or b) a draw from a triangular
         distribution in the interval [-1,1].  Below this is $\bm{\varepsilon}$,
         and in each case, these integrate to 1.
       \item Generate a resampled value of maternal mortality as:  $MMR_t^{b*}=MMR_{ct}+\bm{\varepsilon}\frac{MMR^{UB}_{ct}-MMR^{LB}_{ct}}{2}$, where $MMR_{ct}^{LB}$ is the lower bound estimate and
         $MMR_{ct}^{UB}$ is the upper bound estimate; ie take the original measure, and
         draw a value from the uncertainty interval centred around this measure.
       \item Estimate the original regression using the resampled data, with the re-resampled
         MMR measure.  This results in an estimate of interest $\widehat\beta^{b*}$
       \item if $b<50$ return to step 1.  Else go to step 6
       \item Calculate the standard error of $\widehat\beta$
         as the standard deviation of $\{\widehat\beta^{1*},\widehat\beta^{2*},\ldots,\widehat\beta^{50*}\}$.
         This replaces the original naive standard error, and similarly a p-value can be
         calculated associated with the null hypothesis of a null impact, analogous to
         the p-value calculated based on a standard regression coefficient.
       \end{enumerate}
      
       This procedure is only necessary when estimating impacts of quotas on maternal mortality, and
       not for women in parliament as we are only adjusting for uncertainty in the dependent
       variable in cases where maternal mortality is used.
       Note that in the above we are re-sampling maternal mortality to provide full
       coverage of the 80\% uncertainty interval, or indeed, to provide greater than full
       coverage in the case of the normal draw, for each country year pair.  In each case,
       the normal or triangular distribution places more weight on the likelihood of
       observing a value of maternal mortality close to the stated estimate, and less
       weight on the likelihood of observing a value in the tails of the distribution.
    
       \renewcommand{\thepage}{A\arabic{page}}
       The above resampling procedure assumes that uncertainty in maternal mortality
       is independent between countries and years.  It may also be the case that
       uncertainty is correlated across years within a country.  When undertaking
       inference robust to uncertainty we thus present p-values associated with a
       range of cases as presented in Table \ref{tab:resample}.  These are:
    
       \begin{enumerate}
       \item Bootstrap: The bootstrap analogue of the original p-value (ie no uncertainty in MMR)
       \item Triangular Correction: Resamples from the MMR uncertainty range from the WHO data (80\% coverage) with a triangular distribution whose minimum and maximum are at the end points of the uncertainty range, and whose center is at the estimate
       \item Triangular Correction by Country: Resamples as above, however now instead of taking uncertainty draws by country and year, takes uncertainty draws only at the level of the country.  This implies that uncertainty with regards to MMR is perfectly correlated within a country over time.  It is the limit case of assuming correlation within a country in uncertainty in MMR measurement.
       \item Normal Correction: Resamples from the MMR uncertainty range assuming a normal distribution, where draws are taken so that the 10\textsuperscript{th}/90\textsuperscript{th} quintile of the normal are at the upper and lower end points of the uncertainty range presented in the WHO data in each case.  This allows for us to sample outside the 80\% confidence bounds presented in the original data, and will be the most demanding of all corrections.
       \item Normal Correction by Country: Resamples as above, however now instead of taking uncertainty draws by country and year, takes uncertainty draws only at the level of the country.
       \end{enumerate}

    In results documented in Table \ref{tab:resample} we first provide p-values associated with a standard clustered bootstrap prior to taking into account uncertainty in MMR measurements. Next we provide two sets of bootstrapped p-values computed assuming either a triangular distribution or a normal distribution for MMR uncertainty intervals. While the end points of the triangular distribution are at the ends of the uncertainty interval, the normal distribution provides coverage outside of the 80\% UI. In each of the two types of distributions we allow for the possibility of either assuming  full correlation in uncertainty by country or not. The corrections by country assume full correlation in uncertainty within a country over time. Where not by country, the estimator assumes no correlation within a country over time. The triangular corrections re-sample from MMR so that coverage respects the full 80\% uncertainty interval suggested by the MMEIG. The normal corrections re-sample so that 80\% of re-samples fall within the 80\% uncertainty interval reported by the MMEIG. Both of these inference procedures are implemented assuming no correlation by country, and then assuming full correlation in uncertainty by country.    


\end{spacing}
\end{document}







